Pile PDF

Contents
Preface vi
1 Introduction 1
2 Intact rock and rock mass 9
3 Characterization of discontinuities in rock 36
4 Deformability and strength of rock 80
5 Site investigation and rock testing 173
6 Axial load capacity of drilled shafts in rock 214
7 Axial deformation of drilled shafts in rock 258
8 Lateral load capacity of drilled shafts in rock 285
9 Lateral deformation of drilled shafts in rock 299
10 Stability of drilled shaft foundations in rock 349
11 Drilled shafts in karstic formations 359
12 Loading test of drilled shafts in rock 372
References 404
Index 423
1
Introduction
1.1 DEFINITION OF DRILLED SHAFTS
A drilled shaft is a deep foundation that is constructed by placing concrete in an

excavated hole. Reinforcing steel can be installed in the excavation, if desired, prior to
placing the concrete. A schematic example of a typical drilled shaft socketed into rock is
shown in Figure 1.1. The drilled shaft can carry both axial and lateral loads. Drilled shafts
are sometimes called bored piles, piers, drilled piers, caissons, drilled caissons, or cast-in-
place piles.
To increase the bearing capacity, drilled shafts are commonly socketed into rock. The
portion of the shaft drilled into rock is referred to as a rock socket. In many cases where
there is no overburden soil, drilled shafts entirely embedded in rock are also used.
Because of the high bearing capacity of rock sockets, the analysis and design of them are
extremely important.
This book discusses both rock sockets and drilled shafts entirely embedded in rock.
1.2 HISTORICAL DEVELOPMENT OF DRILLED SHAFTS
The ancient “well foundation” can be considered the earliest version of drilled shafts.
Such foundations were stone masonry pedestals built in hand-excavated holes, long
before hydraulic cements came into common use.
During the late nineteenth and early twentieth centuries when taller and heavier
buildings began to appear, high-capacity foundations became necessary in large cities
such as Chicago, Cleveland and Detroit. These cities are underlain by relatively thick
deposits of medium to soft clays overlying deep glacial till or bedrock. Because
traditional spread foot foundations settled excessively under the heavier building load,
engineers began to use shafts such as the hand-dug “Chicago” and “Gow” caissons. The
shafts were constructed by making the excavation and by placing sections of permanent
liners (wooden lagging or steel rings) to retain the soil by hand.
Hand excavation methods were slow and tedious, so machine-drilled shafts soon
superseded the hand-dug caissons. A few examples of horse and engine-driven augers
appeared between 1900 and 1930, but they had limited capabilities. By the late 1920s,
manufacturers were building practical truck-mounted engine-driven augers, thus bringing
drilled shaft construction into its maturity.
Drilled shafts in rock 2
Fig. 1.1 A drilled shaft socketed into

rock.
During the next three decades, manufacturers and contractors developed larger and more
powerful drilling equipment, which allowed more economical and faster construction of
drilled shafts. In the late 1940s and early 1950s, drilling contractors introduced
techniques for drilling in rock. By introducing casing and drilling mud into boreholes, a
process long established by the oil industry, boreholes could be drilled through difficult
soils economically. By the 1960s, drilled shafts had become a strong competitor to driven
piles.
In the past decade, the use of drilled shafts has increased dramatically. In 1997, the
value of drilled shaft construction in the United States reached more than one billion US
dollars (O’Neill, 1998). Today, drilled shafts support different structures including one-
Introduction 3
story wood frame buildings to the largest skyscrapers, highway bridges, and retaining
structures.
1.3 USE OF DRILLED SHAFTS
Compared to other types of deep foundations, drilled shafts have the following major
advantages:
1. The costs of mobilizing and demobilizing a drill rig are much less than those for a pile
driver.
2. The construction process generates less vibration and noise, making drilled shafts
appropriate for urban construction.
3. The quality of the bearing material can be inspected visually and tests can be run to
determine its physical properties. For end-bearing designs, the soil/rock beneath the
base can be probed for cavities or weak layers if desirable.
4. The diameter or length of the drilled shaft can be easily changed during construction to
compensate for unanticipated soil/rock conditions.
5. The drilled shafts can penetrate through soils with cobbles or boulders. They can also
be socketed into rock.
6. It is usually possible to support very large loads with one large drilled shaft instead of
several piles, thus eliminating the need for a pile cap.
7. Large-diameter drilled shafts are particularly well-suited as foundations for structures
that must resist extreme events that produce large lateral loads (e.g. earthquake and
vessel impact loading) because of the very large moments of inertia.
Drilled shafts also have the following major disadvantages:
1. The quality and performance of drilled shafts is very dependent on the contractor’s
skills. Poor workmanship can produce weak foundations that may not be able to
support the design load.
2. Since shaft construction removes soil/rock from the ground, it may decrease the
competency of the bearing stratum.
3. The construction of drilled shafts through contaminated soils/rocks is problematic
because of the expenses associated with disposing of the spoil.
Because of the above advantages, drilled shafts have become an appropriate and
economical foundation system for heavily loaded structures. When deep foundations are
required, drilled shafts should always be considered as an option.
An application example of drilled shafts in rock

O’Neill and Reese (1999) presented an application example of drilled shafts in rock. It
clearly shows the advantages of drilled shafts over pile-footings for the foundations of the
interior bents of a river bridge in the United States.
The bridge is a two-lane bridge with four spans. Siltstone near the surface at one end
dips to a depth of about 6.1 m (20 ft) near the other end of the bridge. Mixed fine
sediments exist above the siltstone. Two alternate foundation designs were considered by
the design agency before the project bid. The first one called for the construction of one
spread footing and two capped groups of steel H-piles for the three interior bents that
were required to be placed in the river. Both the spread footing and driven piles (with pile
caps) were to be constructed within cofferdams because of the need to construct
footings/caps. The second one called for the replacement of the spread footing and driven
pile groups by three large-diameter drilled shafts. The drilled shafts could be drilled
during low water using a crane-mounted drill rig positioned on timber mats within the
river and pouring the concrete for the shafts to an elevation above the water level,
eliminating the need for cofferdams. Comparison of the pile-footing alternate with the
drilled shaft alternate is shown in Table 1.1. The cost savings realized by using drilled
shafts were $422,000 (50%).
Table 1.1 Comparison of the pile-footing alternate
with the drilled shaft alternate—Queens River
Bridge, Olympic Peninsula, Washington, USA
(after O’Neill & Reese, 1999).
Pile-Footings Drilled Shafts
Details 25 capped H-piles driven into the soft Three 3.2 m (10.5 ft) diameter drilled
siltstone for each of the two interior bents shafts socketed about 10m (30 ft) into the
and a spread-footing at the other interior siltstone, with casing extending from the
bent. All pile driving, cap construction and top of the siltstone to high water level.
spread footing construction were within The casing was used as a form, and the
cofferdams. A single-bent column was drilled shaft concrete was poured directly
formed on top of the spread footing or the up to the top of the casing. The single
pile cap prior to removal of the cofferdams. columns for the bents were formed on top
The construction of work trestle was of the extended sections of drilled shafts,
required so that cofferdams could be with no requirement to construct
constructed prior to installing the cofferdams.
foundations. Because of the length of time
required to construct the trestle and
cofferdams, construction of pile groups,
caps and footing could not proceed until the
following working season, since operations
in the river had to be suspended during the
salmon runs.
Estimated $842,000 $420,000
Cost
1.4 CHARACTERISTICS OF DRILLED SHAFTS IN ROCK
The characteristics of drilled shafts in rock are closely related to the special properties of
rock masses. The following briefly describes some of the special rock mass properties
that will affect the performance of drilled shafts.
Introduction 5
1.4.1 Effect of discontinuities

The primary difference between drilled shafts in rock and those in soil is that rock masses
contain discontinuities. The intact rock may have a high strength but the presence of
discontinuities in the rock may result in very low strength of the rock mass. Wedges or
blocks formed by sets of unfavorably orientated discontinuities may fail by sliding or
toppling, causing excessive movement or failure of drilled shaft foundations.
Figure 1.2 shows the drilled shaft foundations of a bridge across a river. The rock at
this site consists of two sets of discontinuities with about the same dip angles; but set A is
discontinuous and more widely spaced than set B. At the East side, the drilled shaft
foundation would be stable because the discontinuities approximately parallel to the rock
slope face are not continuous. In contrast, at the West side, the discontinuities
approximately parallel to and dipping out of the slope face are continuous and movement
of the entire foundation along these discontinuities is possible.
Fig. 1.2 Effect of discontinuities on the

stability of drilled shafts.
1.4.2 Effect of groundwater

Groundwater may affect the performance of drilled shafts in the following ways:
1. The most obvious is through the operation of the effective stress law. Water under
pressure in the discontinuities defining rock blocks reduces the normal effective stress
between the rock surfaces, and thus reduces the potential shear resistance which can
be mobilized by friction. In porous rocks, such as sandstones, the effective stress law
is obeyed as in granular soils. In both cases, the effect of fissure or pore water pressure
is to reduce the ultimate strength of the rock mass, and thus decrease the bearing
capacity of the drilled shaft foundation.
2. Groundwater affects rock mechanical properties due to the deleterious action of water
on particular rocks and minerals. For example, clay seams may soften in the presence
of groundwater, reducing the strength and increasing the deformability of the rock
mass. Argillaceous rocks, such as shales and argillitic sandstones, also demonstrate
marked reductions in material strength following infusion with water. According to
Hoek and Brown (1997), strength losses of 30–100% may occur in many rocks as a
result of chemical deterioration of the cement or clay binder.
3. Groundwater flow into the excavation of a drilled shaft can make cleaning and
inspection of bearing surfaces difficult and result in decreased bearing capacity for the
drilled shaft.
1.4.3 Effect of karstic formations

A number of problems may arise when drilled shafts are built in karstic formations
(Brown, 1990; Goodman, 1993; Sowers, 1996):
1. An existing cavity may underlie the base of the drilled shaft and collapse when the
building is under construction or in service. The collapse of the cavity may be caused
by excessive construction loading or erosion by acid groundwater [Fig. 1.3(a)].
2. The tip of the shaft may slide along a steeply inclined rock instead of penetrating into
bedrock, especially when part of the tip is located on top of a pinnacle with existing
joints or cracks [Fig. 1.3(b)].
3. The drilled shaft is placed on a cantilever rock over cavities or soft clay, so that
excessive loads or continuing water erosion may cause rock collapse [Fig. 1.3(c)].
4. A shifting rock slab or rock block floating in the residual soil may lead people to
mistakenly believe that the bedrock has been reached and the bearing stratum has been
located (d)].
Chapter 11 will discuss the performance of drilled shafts in karstic formations in more
detail.
1.5 CONSIDERATIONS IN THE DESIGN OF DRILLED SHAFTS
As for the design of any foundations, the design of drilled shafts must satisfy criteria
related to strength, deformation and durability. For the strength, criteria are applied to
both the structural strength of the shaft itself and the geotechnical strength, i.e., the load
carrying capacity of the soil/rock. The structural and geotechnical strength criteria depend
on the basis of the design method. The traditional working stress design method,
sometimes referred to as the allowable stress design (ASD) method, relies on an overall
safety factor against ultimate failure and the corresponding design criteria can be
expressed as
(1.1)
Introduction 7
where Qu is the ultimate load bearing capacity; FS is the global factor of safety; and Q is
the allowable working load or the allowable design load.
Equation (1.1) applies to both axial and lateral loadings. Typical factors of safety for
the geotechnical strength of drilled shafts range between two and three, depending on the
method of capacity calculation, the extent of the designer’s experience and knowledge of
the site and the geotechnical conditions, and the likely consequences of failure. In cases
where there is extensive experience of the site and field shaft load tests have been carried
out, values of safety factor as low as 1.5 may be appropriate. On the other hand, where
knowledge of the site is limited, and the consequences of failure may be extreme, safety
factors of three or higher may be appropriate.
Fig. 1.3 Failure modes of drilled shafts

in karstic formations (after Tang,
1995).
In recent years, there has been an increasing tendency to use load and resistance factor
design (LRFD) for drilled shafts and other structural components (AASHTO, 1994;
FHWA, 1996a). With this method, various factors, with values of 1 or above, are applied
to the individual components of load. Other factors, with values of 1 or less, are applied
to the total resistance, or individual components of resistance, in such a way to assure a
margin of safety consistent with historical practice using global factors of safety. The
design criterion for the LRFD approach can be written as
(1.2)
where η is factor varying from 0.95 to 1.05 to reflect ductility, redundancy and
operational importance of the structure; γi is the load factor for load type i; Qi is the
nominal value of load type i; is the resistance factor for resistance component j; Quj is
the estimated (nominal) value of ultimate resistance component j.
Equation (1.2) applies to both axial and lateral loadings, and to structural and
geotechnical strengths. The LRFD approach to foundation design has the advantages that
(a) foundations are easier to design if the superstructure is designed using LRFD and (b)
it offers a means to incorporate reliability into the design process in a rational manner.
For the serviceability limit state, the design criteria for deformations may be stated
generally as:
Estimated deformation≤Allowable deformation
(1.3)
Estimated differential deformation≤Allowable differential
deformation (1.4)
Equations (1.3) and (1.4) apply to both axial and lateral deformations. The allowable
deformations and differential deformations depend primarily on the nature of the
structure. Grant et al. (1982) and Moulton et al. (1985) listed typical values of allowable
deformations and differential deformations for different structures.
For the durability, the usual design criterion is the drilled shafts shall have a design
life that exceeds the design life of the structure to be supported; this is usually 50 years or
more for permanent structures.
In recent years, the influence of environmental factors on the design and construction
of drilled shaft foundations has become more and more important. Requirements that
impact the excavation, handling, and disposal of river bottom sediments are continually
more restrictive. Consequently, design and construction techniques are being developed
and modified to lessen the need for excavation.
2
Intact rock and rock mass
2.1 INTRODUCTION
Rock differs from most other engineering materials in that it contains discontinuities of
one type or another which render its structure discontinuous. Thus a clear distinction
must be made between the intact rock or rock material on the one hand and the rock mass
on the other. The intact rock may be considered as a continuum or polycrystalline solid
between discontinuities consisting of an aggregate of minerals or grains. The rock mass is
the in situ medium comprised of intact rock blocks separated by discontinuities such as
joints, bedding planes, folds, sheared zones and faults. The properties of the intact rock
are governed by the physical properties of the materials of which it is composed and the
manner in which they are bonded to each other. The parameters which may be used in a
description of intact rock include petrological name, color, texture, grain size, minor
lithological characteristics, degree of weathering or alternation, density, porosity,
strength, hardness and deformability. Rock masses are discontinuous and often have
heterogeneous and anisotropic properties. Since the behavior of a rock mass is, to a large
extent, determined by the type, spacing, orientation and characteristics of the
discontinuities present, the parameters used to describe a rock mass include the nature
and geometry of discontinuities as well as its overall strength and deformability.
This chapter describes the types and important properties of intact rocks and different
rock mass classification systems that will be useful in the analysis and design of drilled
shafts in rock. Chapter 3 will discuss the characterization of discontinuities in rock
masses.
2.2 INTACT ROCK
Intact rocks may be classified from a geological or an engineering point of view. In the
first case the mineral content of the rock is of prime importance, as is its texture and any
change which has occurred since its formation. Although geological classifications of
intact rocks usually have a genetic basis, they may provide little information relating to
the engineering behavior of the rocks concerned since intact rocks of the same geological
category may show a large scatter in strength and deformability, say of the order of 10
times. Therefore, engineering classifications of intact rocks are more related to the
analysis and design of foundations in rock.
2.2.1 Geological classification
(a) Rock-forming minerals

Rocks are composed of minerals, which are formed by the combination of naturally
occurring elements. Although there are hundreds of recognized minerals, only a few are
common. Table 2.1 summarizes the common rock-forming minerals and their properties.
Moh’s scale, used in the table, is a standard of ten minerals by which the hardness of a
mineral may be determined. Hardness is defined as the ability of a mineral to scratch
another. The scale is one for the softest mineral (talc) and ten for the hardest (diamond).
Table 2.1 Common rock-forming minerals and their
properties.
Mineral Hardness Relative Fracture Structure
(Moh’s scale, Density
1–10)
Orthoclase 6 2.6 Good cleavage at right Monoclinic. Commonly
feldspar angles occurs as crystals
Plagioclase 6 2.7 Cleavage nearly at right Triclinic. Showing
feldspar angles—very marked distinct cleavage
lamellae
Quartz 7 2.65 No cleavage. hexagonal
Choncoidal fracture
Muscovite 2.5 2.8 Perfect single cleavage Monoclinic. Exhibiting
into thin easily separated strong cleavage
plates lamellae
Biotite 3 3 Perfect single cleavage Monoclinic. Exhibiting
into thin easily separated strong cleavage
plates lamellae
Hornblende 5–6 3.05 Good cleavage at 120° Hexagonal—normally
in elongated prisms
Augite 5–6 3.05 Cleavage nearly at right Monoclinic
angles
Olivine 6–7 3.5 No cleavage No distinctive structure
Calcite 3 2.7 Three perfect cleavages. Hexagonal
Rhomboids formed
Dolomite 4 2.8 Three perfect cleavages Hexagonal
Kaolinite 1 2.6 No cleavage No distinctive structure
(altered feldspar)
Hematite 6 5 No cleavage Hexagonal
Intact rock and rock mass 11
(b) Elementary rock classification

Intact rocks are classified into three main groups according to the process by which they
are formed: igneous, sedimentary and metamorphic.
Igneous rocks are formed by crystallization of molten magma. The mode of
crystallization of the magma, at depth in the Earth’s crust or by extrusion, and the rate of
cooling affect the rock texture or crystal size. The igneous rocks are subdivided into
plutonic, hypabyssal and extrusive (volcanic), according to their texture. They are further
subdivided into acid, intermediate, basic and ultrabasic, according to their silica content.
Table 2.2 shows a schematic classification of igneous rocks.
Sedimentary rocks are formed from the consolidation of sediments. Sedimentary rocks
cover three-quarters of the continental areas and most of the sea floor. In the process of
erosion, rocks weather and are broken down into small particles or totally dissolved.
These detritic particles may be carried away by water, wind or glaciers, and deposited far
from their original position. When these sediments start to form thick deposits, they
consolidate under their own weight and eventually turn into solid rock through chemical
or biochemical precipitation or organic process. As a result of this process, sedimentary
rocks almost invariably possess a distinct stratified, or bedded, structure. Table 2.3 shows
the classification of sedimentary rocks.
Metamorphic rocks are derived from pre-existing rocks by temperature, pressure
and/or chemical changes. Table 2.4 shows a classification of the metamorphic rocks
according to their physical structure, i.e., massive or foliated.
(c) Weathering of rock

Weathering is the disintegration and decomposition of the in situ rock, which is generally
depth controlled, that is, the degree of weathering decreases with increasing depth below
the surface. The engineering properties of a rock as discussed in next section can be, and
often are, altered to varying degrees by weathering of the rock material. Intact rocks can
be divided into 5 groups according to the degree of weathering (see Table 2.5).
Table 2.2 Geological classification of igneous
rocks.
Type
Grain size Acid >65% silica Intermediate 55– Basic 45–55% Ultrabasic
65% silica silica <45% silica
Plutonic Granite Diorite Gabbro Picrite
(coarse) Granodiorite Peridotite
Serpentinite
Dunite
Hypabyssal Quartz Plagioclase Dolerite Basic dolerites
Orthoclase porphyries porphyries
Extrusive Rhyolite Pichstone Basalt Basic olivine
(fine) Dacite Andesite basalts
Major mineral Quartz, orthoclase, Quartz, orthoclase, Calcium- Calcium-

constituents sodium-plagioclase, plagioclase, biotite, plagioclase, plagioclase,
muscovite, biotite, hornblende, augite augite, olivine, olivine, augite
hornblende hornblende
Table 2.3 Classification of sedimentary rocks.

Method of Classification Rock Description Major mineral
formation constituents
Mechanical Rudaceous Breccia Large grains in clay Various
Conglomerate matrix
Arenaceous Sandstone Medium, round Quartz, calcite
grains in calcite (sometimes
matrix feldspar, mica)
Quartzite Medium, round Quartz
grains in silica matrix
Gritstone Medium, angular Quartz, calcite,
grains in matrix various
Breccia Coarse, angular Quartz, calcite,
grains in matrix various
Argillaceous Claystone Micro-fine-grained Kaolinite, quartz,
plastic texture mica
Shale Harder-laminated Kaolinite, quartz,
Mudstone compacted clay mica
Organic Calcareous Limestone Fossiliferous, coarse Calcite
or fine grained
Carbonaceous Coal
(siliceous,
ferruginous,
phosphatic)
Chemical Ferruginous Ironstone Impregnated Calcite, iron oxide
limestone or
claystone (or
precipitated)
Calcareous (siliceous, Dolomite Precipitated or Dolomite, calcite
saline) limestone replaced limestone,
fine grained
Table 2.4 Classification of metamorphic rocks.

Classification Rock Description Major mineral
constituents
Massive Hornfels Micro-fine grained Quartz
Quartzite Fined grained Quartz

Marble Fine—coarse grained Calcite or dolomite
Foliated Slate Micro-fine grained, laminated Kaolinite, mica
Phyllite Soft, laminated Mica, kaolinite
Schist Altered hypabyssal rocks, coarse Feldspar, quartz, mica
grained
Gneiss Altered granite Hornblende
Table 2.5 Degree of weathering of intact rocks.

Degree of Description
weathering
Unweathered No evidence of any chemical or mechanical alteration
Slightly Slight discoloration on surface, slight alteration along discontinuities, less than
weathered 10 percent of the rock volume altered
Moderately Discoloring evident, surface pitted and altered with alteration penetrating well
weathered below rock surfaces, weathering “halos” evident, 10 to 50 percent of the rock
altered
Highly weathered Entire mass discolored, alteration pervading nearly all of the rock with some
pockets of slightly weathered rock noticeable, some minerals leached away
Decomposed Rock reduced to a soil with relict rock texture, generally molded and crumbled
by hand
Table 2.6 Engineering classification of rock by

strength (from ISRM, 1978; CGS, 1985; Marinos &
Hoek, 2001).
Grade Classification Field identification Unconfined Point Examples
compressive Load
strength Index
(MPa) (MPa)
R0 Extremely Indented by thumbnail <1 –1) Stiff fault gouge
weak
R1 Very weak Crumbles under firm 1–5 –1) Highly weathered
blows of geological or altered rock,
hammer; can be peeled shale
with a pocket knife
R2 Weak Can be peeled with a 5–25 –1) Chalk, claystone,
pocket knife with potash, marl,
difficulty; shallow siltstone, shale,
indentations made by a rock salt
firm blow with point of
geological hammer
R3 Medium strong Cannot be scraped or 25–50 1–2 Concrete, phyllite,
peeled with a pocket schist, siltstone
knife; specimen can be
fractured with a single
firm blow of geological
hammer
R4 Strong Specimen requires more 50–100 2–4 Limestone, marble,
than one blow of sandstone, schist
geological hammer to
fracture
R5 Very strong Specimen requires 100–250 4–10 Amphibolite,
many blows of sandstone, basalt,
geological hammer to gabbro, gneiss,
fracture granodiorite,
peridotite, rhyolite,
tiff
R6 Extremely Specimen can only be >250 >10 Fresh basalt, chert,
strong chipped with the diabase, gneiss,
geological hammer granite, quartzite
1)
Point load tests on rocks with unconfined compressive strength below 25 MPa are likely to yield
highly ambiguous results.
2.2.2 Engineering classification

The engineering classification of intact rocks is based on strength and/or deformation
properties of the rock. Table 2.6 shows the classification system of the International
Society of Rock Mechanics (ISRM, 1978). The ISRM classification is also recommended
in the Canadian Foundation Engineering Manual (CGS, 1985). In this classification, the
rock may range from extremely weak to extremely strong depending on the unconfined
compressive strength or the approximate field identification.
Based on laboratory measurements of strength and deformation properties of rocks,
Deere and Miller (1966) established a classification system based on the ultimate strength
(unconfined compressive strength) and the tangent modulus Et of elasticity at 50% of the
ultimate strength. Figure 2.1 summarizes the engineering classification of igneous,
sedimentary and metamorphic rocks, respectively, as given in Deere and Miller (1966).
The modulus ratio in these figures is that of the elastic modulus to the unconfined
compressive strength. A rock may be classified as AM, BH, BL, etc. Voight (1968),
however, argued that the elastic properties of intact rock could be omitted from practical
classification since the elastic moduli as determined in the laboratory are seldom those
required for engineering analysis.
2.2.3 Typical values of intact rock properties

This section lists the typical values of intact rock properties, including density (Table
2.7), unconfined compressive strength (Table 2.8), elastic modulus (Table 2.9) and
Poisson’s ratio (Table 2.10). These values are listed only for reference and should not be
used directly in design.
2.3 ROCK MASS
Numerous rock mass classification systems have been developed, including Terzaghi’s
Rock Load Height Classification (Terzaghi, 1946); Lauffer’s Classification (Lauffer,
1958); Deere’s Rock Quality Designation (RQD) (Deere, 1964); RSR Concept (Wickham
et al., 1972); the Rock Mass Rating (RMR) system (Bieniawski, 1973, 1976, 1989); the
Q-System (Barton et al., 1974), and the Geological Strength Index (GSI) system (Hoek &
Brown, 1997). Most of the above systems were primarily developed for the design of
underground excavations. However, four of the above classification systems have been
used extensively in correlation with parameters applicable to the design of rock
foundations. These four classification systems are the Rock Quality Designation (RQD),
the Rock Mass Rating (RMR), the Q-System, and the Geological Strength Index (GSI).
2.3.1 Rock quality designation (RQD)
Rock Quality Designation (RQD) was introduced by Deere (1964) as an index assessing
rock quality quantitatively. The RQD is defined as the ratio (in percent) of the total length
of sound core pieces 4 in. (10.16 cm) in length or longer to the length of the core run.
RQD is perhaps the most commonly used method for characterizing the jointing in
borehole cores, although this parameter may also implicitly include other rock mass
features such as weathering and core loss. Deere (1964) proposed the relationship
between the RQD index and the rock mass quality as shown in Table 2.11.
(a) Direct method for determining RQD

For RQD determination, the International Society for Rock Mechanics (ISRM)
recommends a core size of at least NX (size 54.7 mm) drilled with double-tube core
barrel using a diamond bit. Artificial fractures can be identified by close fitting of cores
and unstained surfaces. All the artificial fractures should be ignored while counting the
core length for RQD. A slow rate of drilling will also give better RQD. The correct
procedure for measuring RQD is shown in Figure 2.2.
Fig. 2.1 Engineering classification of

intact rocks (Et is the tangent modulus
at 50% ultimate strength) (after Deere
& Miller, 1966).
Table 2.7 Typical values of density of intact rocks

(after Lama & Vutukuri, 1978).
Rock type Range of density (kg/m3) Mean density (kg/m3)
Igneous rocks
Granite 2516–2809 2667
Granodiorite 2668–2785 2716
Syenite 2630–2899 2757
Quartz diorite 2680–2960 2806
Diorite 2721–2960 2839
Norite 2720–3020 2984
Gabbro 2850–3120 2976
Diabase 2804–110 2965
Peridotite 3152–3276 3234
Dunite 3204–3314 3277
Pyroxenite 3100–3318 3231
Anorthosite 2640–2920 2734
Sedimentary rocks
Sandstone 2170–2700 –
Limestone 2370–750 –
Dolomite 2750–2800 –
Chalk 2230 –
Marble 2750 –
Shale 2060–2660 –
Sand 1920–1930 –
Metamorphic rocks
Gneiss 2590–3060 2703
Schist 2700–3030 2790
Slate 2720–840 2810
Amphibolite 2790–3140 2990
Granulite 2630–3100 2830
Eclogite 3338–3452 3392
Note: The values listed in the table are for the bulk density determined at natural water content.
Table 2.8 Typical range of unconfined compressive

strength of intact rocks (AASHTO, 1989).
Rock General description Rock Unconfined compressive
category strength, σc(1) (MPa)
A Carbonate rocks with well- Dolostone 33–310
developed crystal cleavage Limestone 24–290
Carbonatite 38–69
Marble 38–241
Tactite-Skarn 131–338
B Lithified argillaceous rock Argillite 29–145
Claystone 1–8
Marlstone 52–193
Phyllite 24–241
Siltstone 10–117
Shale(2) 7–35
Slate 145–207
C Arenaceous rocks with strong Conglomerate 33–221
crystals and poor cleavage Sandstone 67–172
Quartzite 62–379
D Fine-grained igneous crystalline Andesite 97–179
rock Diabase 21–572
E Coarse-grained igneous and Amphibolite 117–276
metamorphic crystalline rock Gabbro 124–310
Gneiss 24–310
Granite 14–338
Quartz diorite 10–97
Quartz 131–159
monozonite 10–145
Schist 179–427
Syenite
(1)
Range of unconfined compressive strength reported by various investigations.
(2)
Not including oil shale.
Table 2.9 Typical values of elastic modulus of

intact rocks (AASHTO, 1989).
No. of No. of rock Elastic modulus (GPa) Standard
values types Deviation
Rock Maximum Minimum Mean
type
Granite 26 26 100 6.41 52.7 24.5
Diorite 3 3 112 17.1 51.4 42.7
Gabbro 3 3 84.1 67.6 75.8 6.69
Diabase 7 7 104 69.0 88.3 12.3

Basalt 12 12 84.1 29.0 56.1 17.9
Quartzite 7 7 88.3 36.5 66.1 16.0
Marble 14 13 73.8 4.00 42.6 17.2
Gneiss 13 13 82.1 28.5 61.1 15.9
Slate 11 2 26.1 2.41 9.58 6.62
Schist 13 12 69.0 5.93 34.3 21.9
Phyllite 3 3 17.3 8.62 11.8 3.93
Sandstone 27 19 39.2 0.62 14.7 8.21
Siltstone 5 5 32.8 2.62 16.5 11.4
Shale 30 14 38.6 0.007 9.79 10.0
Limestone 30 30 89.6 4.48 39.3 25.7
Dolostone 17 16 78.6 5.72 29.1 23.7
Table 2.10 Typical values of Poisson’s ratio of

intact rocks (AASHTO, 1989).
No. of No. of rock Poisson’s ratio Standard
values types Deviation
Rock Maximum Minimum Mean
type
Granite 22 22 0.39 0.09 0.20 0.08
Gabbro 3 3 0.20 0.16 0.18 0.02
Diabase 6 6 0.38 0.20 0.29 0.06
Basalt 11 11 0.32 0.16 0.23 0.05
Quartzite 6 6 0.22 0.08 0.14 0.05
Marble 5 5 0.40 0.17 0.28 0.08
Gneiss 11 11 0.40 0.09 0.22 0.09
Schist 12 11 0.31 0.02 0.12 0.08
Sandstone 12 9 0.46 0.08 0.20 0.11
Siltstone 3 3 0.23 0.09 0.18 0.06
Shale 3 3 0.18 0.03 0.09 0.06
Table 2.11 Correlation between RQD and rock

mass quality.
RQD (%) Rock Mass Quality
<25 Very poor
25–50 Poor
50–75 Fair
75–90 Good
90–100 Excellent
(b) Indirect methods for determining RQD

The seismic survey method can be used to determine the RQD indirectly. By comparing
the in situ compressional wave velocity with laboratory sonic velocity of intact drill core
obtained from the same rock mass, the RQD can be estimated by
RQD(%)=Velocity ratio=(VF/VL)2×100
(2.1)
where VF is the in situ compressional wave velocity; and VL is the compressional wave
velocity in intact rock core.
When cores are not available, RQD may be estimated from the number of joints
(discontinuities) per unit volume JV. A simple relationship which may be used to convert
JV into RQD for clay-free rock masses is (Palmstrom, 1982)
RQD(%)=115−3.3JV
(2.2)
where JV is the total number of joints per cubic meter or the volumetric joint count.
The volumetric joint count JV has been described by Palmstrom (1982, 1985, 1986) and
Sen and Eissa (1992). It is a measure for the number of joints within a unit volume of
rock mass defined by
(2.3)
Fig. 2.2 Procedure for measurement

and calculation of rock quality
designation RQD (after Deere, 1989).
where si is the average joint spacing in meters for the ith joint set and N is the total
number of joint sets except the random joint set.
Random joints may also be considered by assuming a “random spacing”. Experience
indicates that this should be set to sr=5 m (Palmstrom, 1996). Thus, the volumetric joint
count can be generally expressed as
(2.4)
where Nr is the total number of random joint sets and can easily be estimated from joint
observations. In cases where random or irregular jointing occurs, JV can be found by
counting all the joints observed in an area of known size.
Though the RQD is a simple and inexpensive index, when considered alone it is not
sufficient to provide an adequate description of a rock mass because it disregards joint
orientation, joint condition, type of joint filling and other features.
2.3.2 Rock mass rating (RMR)

The Geomechanics Classification, or Rock Mass Rating (RMR) system, proposed by
Bieniawski (1973), was initially developed for tunnels. In recent years, it has been
applied to the preliminary design of rock slopes and foundations as well as for estimating
the in-situ modulus of deformation and rock mass strength. The RMR uses six parameters
that are readily determined in the field (see Table 2.12):
• Unconfined compressive strength of the intact rock
• Rock Quality Designation (RQD)
• Spacing of discontinuities
• Condition of discontinuities
• Ground water conditions
• Orientation of discontinuities
All but the intact rock strength are normally determined in the standard geological
investigations and are entered on an input data sheet. The guidelines for assessing the
discontinuity condition is shown in Table 2.13. The unconfined compressive strength of
rock is determined in accordance with standard laboratory procedures but can be readily
estimated in situ from the point-load strength index (see Table 2.12).
Rating adjustments for discontinuity orientation are summarized for underground
excavations, foundations and slopes in Part B of Table 2.12. A more detailed explanation
of these rating adjustments for dam foundations is given in Table 2.14, after ASCE
(1996).
The six separate ratings are summed to give an overall RMR, with a higher RMR
indicating better quality rock. Based on the observed RMR value, the rock mass is
classified into five classes named as very good, good, fair, poor and very poor, as shown
in Part C of Table 2.12. Also shown in Part C of Table 2.12 is an interpretation of these
five classes in terms of roof stand-up time, cohesion, internal friction angle and
deformation modulus for the rock mass.
2.3.3 Rock mass quality (Q)

The Q-system, proposed by Barton et al. (1974), was developed specifically for the
design of tunnel support systems. As in the case of the Geomechanics System, the Q-
system has been expanded to provide preliminary estimates of rock mass properties.
Likewise, the Q-system incorporates the following six parameters and the equation for
obtaining rock mass quality Q:
Table 2.12 Geomechanics classification of jointed

rock masses (after Bieniawski, 1976, 1989).
A. Classification parameters and their rating
Parameter Range of values
1 Strength Point-load >10 4–10 2–4 1–2 For this low
of intact strength range,
rock index (MPa) unconfined
compressive
test is
preferred
Unconfined >250 100–250 50–100 25–50 5– 1– <1
compressive 25 5
strength
(MPa)
Rating 15 12 7 4 2 1 0
2 Drill core quality RQD 90–100 75–90 50–75 25–50 <25
(%)
Rating 20 17 13 8 3
3 Spacing of >2 0.6–2 0.2–0.6 0.06–0.2 <0.06
discontinuities (m)
Rating 20 15 10 8 5
4 Conditions of Very rough Slightly Slightly Slickensided Soft gouge
discontinuities surfaces, Not rough rough surfaces or >5 mm thick
continuous, surfaces, surfaces, Gouge <5 or Separation
No separation, separation separation mm thick or >5 mm
Unweathered <1 mm, <1 mm, Separation 1– Continuous
wall rock Slightly Highly 5 mm
weathered weathered continuous
walls walls
Rating 30 25 20 10 0
5 Ground Inflow per None or <10 or 10–25 or 25–125 or >125 or
water 10m tunnel
length
(1/min)
Ratio of 0 or <0.1 or 0.1–0.2 or 0.2–0.5 or >0.5 or
joint water
pressure to
major
principal
stress
General Completely Damp Wet Dripping Flowing
conditions dry
24
Rating 15 10 7 4 0
B. Rating adjustment for joint orientations

Strike and dip orientations of Very Favorable Fair Unfavorable Very
discontinuities favorable Unfavorable
Tunnels and mines 0 −2 −5 −10 −12
Ratings Foundations 0 −2 −7 −15 −25
Slopes 0 −5 −25 −50 −60
C. Rock mass classes and corresponding design parameters and engineering properties
Class No. I II III IV V
RMR 100→81 80→61 60→41 40→21 <20
Description Very Good Good Fair Poor Very poor
Average stand-up time 20 years for 1 year for 1 week for 10 hours for 30 minutes for
15 m span 10m span 5m span 2.5 m span 1 m span
Cohesion of rock mass >0.4 0.3–0.4 0.2–0.3 0.1–0.2 <0.1
(MPa)
Internal friction angle >45 35–45 25–35 15–25 <15
of rock mass (°)
Deformation modulus >56 56–18 18–5.6 5.6–1.8 <1.8
(GPa)a)
a)
Deformation modulus values are from Serafim and Pereira (1983).
Table 2.13 Guidelines for classifying discontinuity

condition (after Bieniawski, 1989).
Parameter Range of values
Discontinuity Rating 6 4 2 1 0
length (persistence/
continuity) Measurement (m) <1 1–3 3–10 10–20 >20
Separation Rating 6 5 4 1 0
(aperture)
Measurement None <0.1 0.1–1 1–5 >5
(mm)
Roughness Rating 6 5 3 1 0
Description Very Rough Slight Smooth Slickensided
rough
Infilling (gouge) Rating 6 4 2 2 0
Description and None Hard Hard Soft Soft filling

Measurement filling filling >5 filling <5 >5
(mm) <5
Degree of Rating 6 5 3 1 0
weathering
Description None Slight Moderate High Decomposed
Note: Some conditions are mutually exclusive. For example, if infilling is present, it is irrelevant
what the roughness may be, since its effect will be overshadowed by the influence of the gouge. In
such cases, use Table 2.12 directly.
Table 2.14 Ratings for discontinuity orientations for

dam foundations (after ASCE, 1996).
Dip 10°–30°
Dip 0°–10° Dip direction Dip 30°–60° Dip 60°–90°
Upstream Downstream
Very favorable Unfavorable Fair Favorable Very unfavorable
• Rock Quality Designation (RQD)

• Number of discontinuity sets
• Roughness of the most unfavorable discontinuity
• Degree of alteration or filling along the weakest discontinuity
• Water inflow
• Stress condition
(2.5)
where RQD=Rock Quality Designation; Jn=joint set number; Jr=joint roughness number;
Ja=joint alteration number; Jw=joint water reduction number; and SRF= stress reduction
number.
The meaning of the parameters used to determine the value of Q in Equation (2.5) can
be seen from the following comments by Barton et al. (1974):
The first quotient (RQD/Jn), representing the structure of the rock mass, is a crude
measure of the block or particle size, with the two extreme values (100/0.5 and 10/20)
differing by a factor of 400. If the quotient is interpreted in units of centimetres, the
extreme ‘particle sizes’ of 200 to 0.5 cm are seen to be crude but fairly realistic
approximations. Probably the largest blocks should be several times this size and the
smallest fragments less than half the size. (Clay particles are of course excluded).
The second quotient (Jr/Ja) represents the roughness and frictional characteristics of
the joint walls or filling materials. This quotient is weighted in favor of rough, unaltered
joints in direct contact. It is to be expected that such surfaces will be close to peak
strength, that they will dilate strongly when sheared, and they will therefore be especially
favorable to tunnel stability.
When rock joints have thin clay mineral coatings and fillings, the strength is reduced
significantly. Nevertheless, rock wall contact after small shear displacements have
occurred may be a very important factor for preserving the excavation from ultimate
failure.
Where no rock wall contact exists, the conditions are extremely unfavorable to tunnel
stability. The ‘friction angles’ (given in Table 2.15) are a little below the residual strength
values for most clays, and are possibly dowti-graded by the fact that these clay bands or
fillings may tend to consolidate during shear, at least if normal consolidation or if
softening and swelling has occurred. The swelling pressure of monttnorillonite may also
be a factor here.
The third quotient (Jw/SRF) consists of two stress parameters. SRF is a measure of: 1)
loosening load in the case of an excavation through shear zones and clay bearing rock, 2)
rock stress in competent rock, and 3) squeezing loads in plastic incompetent rocks. It can
be regarded as a total stress parameter. The parameter Jw is a measure of water pressure,
which has an adverse effect on the shear strength of joints due to a reduction in effective
normal stress. Water may, in addition, cause softening and possible out-wash in the case
of clay-filled joints. It has proved impossible to combine these two parameters in terms of
inter-block effective stress, because paradoxically a high value of effective normal stress
may sometimes signify less stable conditions than a low value, despite the higher shear
strength. The quotient (Jw/SRF) is a complicated empirical factor describing the ‘active
stress’.
So the rock mass quality (Q) may be considered a function of three parameters which
are approximate measures of:
(i) Block size (RQD/Jn): It represents the overall structure of rock masses.
(ii) Inter block shear strength (Jr/Ja): It represents the roughness and frictional
characteristics of the joint walls or filling materials.
(iii) Active stress (Jw/SRF): It is an empirical factor describing the active stress.
Table 2.15 provides the necessary guidance for assigning values to the six parameters.
Depending on the six assigned parameter values reflecting the rock mass quality, Q can
vary between 0.001 to 1000. Rock quality is divided into nine classes ranging from
exceptionally poor (Q ranging from 0.001 to 0.01) to exceptionally good (Q ranging from
400 to 1000) as shown in Table 2.16.
Since the Q and RMR systems are based on much the same properties, they are highly
correlated and can be predicted one from the other. Various authors give a relationship in
the following form (Rutledge & Preston, 1978; Cameron-Clarke & Budavari, 1981; Abad
et al., 1984; Beniawski, 1989; Goel et al., 1995):
Table 2.15 The Q-system and associated parameters

RQD, Jn, Jr, Ja, SRF and Jw (after Barton et al.,
1974).
Rock Quality
Designation
RQD (%)
Very poor 0–25 Note:
Poor 25–0 (i) Where RQD is reported or measured to be
<10 a nominal value of 10 is used to
Fair 50–5 evaluate Q in Equation (2.5)
Good 75–90
Excellent 90–100
(ii) Take RQD to be nearest 5%

Joint Set
Number
Jn
Massive, none or few joints 0.5–1.0 Note:
One joint set 2 (i) For intersections use (3.0×Jn)
One joint set plus random 3
Two joint sets 4 (ii) For portals use (2.0×Jn)
Two joint sets plus random 6
Three joint sets 9
Three joint sets plus random 12
Four or more joint sets, 15
random, heavily jointed,
‘sugar cube’, etc
Crushed rock, earthlike 20
Joint
Roughness
Number
Jr
(a) Rock wall contact and Notes:
(b) Rock wall contact (i) Add 1.0 if the mean spacing of the relevant
joint set is greater than 3 m
before 10 cm shear
Discontinuous joint 4
Rough or irregular, 3 (ii) Jr=0.5 can be used for planar slickensided
joints having lineations, provided the
undulating lineations are favorably orientated
Smooth, undulating 2
Slickensided, undulating 1.5
Rough and irregular, planar 1.5
Smooth or irregular 1
Slickensided, planar 0.5
(c) No rock wall contact
when sheared
Zone containing clay 1 (nominal)
minerals thick enough to
prevent rock wall contact
Sandy, gravelly or crushed 1 (nominal)
zone thick enough to
prevent rock wall contact
Joint Approximate residual angle of friction
Alternation (deg)
Number
Ja
(a) Rock wall contact
A. Tightly healed, hard, non- 0.75 –
softening,
impermeable filling, i.e. quartz or epidote

B. Unaltered joint walls, surface staining only 1 25–
35
C. Unlatered joint walls. Non-softening mineral coatings, sandy particles, clay-free 2 25–
disintegrated rock, etc. 30
D. Silty or sandy clay coatings, small clay fraction (non-softening) 3 20–
25
E. Softening or low friction clay mineral coatings, i.e. kaolinite, mica. Also chlorite, 4 8–
talc, gypsum and graphite, etc, and small quantities of swelling clays 16
(discontinuous coatings, 1–2 mm or less in thickness)
(b) Rock wall contact before 10 cm shear
F. Sandy particles, clay free disintegrated rock, etc 4 25–
30
G. Strongly over-consolidated, non-softening clay mineral fillings (continuous, <5 6 16–
24
H. Medium or low over-consolidation, softening, clay mineral fillings (continuous, 8 12–
<5 mm in thickness) 16
J. Swelling clay fillings, i.e. montmorillonite (continuous, <5 mm in thickness). 8–12 6–
Value of Ja depends on percentage of swelling clay-sized particles and access to 12
water, etc
(c) No rock wall contact when sheared
K. Zones or bands of disintegrated or crushed 6, 8 6–
or 8– 24
12
rock and clay (see G, H. J for

description of clay condition)
L. Zones or bands of silty or sandy 5 –
clay, small clay fraction (non-
softening)
M. Thick, continuous zones or bands of 10, 13 or 13– 6–24
clay (see G, H. J for description of 20
clay condition)
Joint Water Approximate water pressure (kPa)
Reduction
Factor
Jw
A. Dry excavations or minor inflow, i.e. 1 <100
<5 l/min locally
B. Medium inflow or pressure 0.66 100–250
occasional outwash of joint fillings
C. Large inflow or high pressure in 0.5 250–1000
competent rock with unfilled joints
D. Large inflow or high pressure, 0.33 250–1000
considerable occasional outwash of
joint fillings
E. Exceptionally high inflow or water 0.1 >1000
pressure at blasting, decaying with
time
F. Exceptionally high inflow or water 0.1−0.05 >1000
pressure continuing without decay
Note:
(i) Factors C-F are crude estimates. Increase Jw if drainage measures are installed
(ii) Special problems caused by ice formation are not considered
Stress
Reduction
Factor SRF
(a) Weakness zones intersecting
excavation, which may cause
loosening of rock mass when tunnel
is excavated
A. Multiple occurrences of 10 Note:
weakness zones containing clay or (i) Reduce these values by 25–50% if
chemically disintegrated rock, very the relevant shear zones only
loose influence but do not intersect the
excavation
surrounding rock (any

depth)
B. Single weakness 5
zones containing clay
or chemically
disintegrated rock
(depth of excavation
<50 m)
C. Single weakness 2.5
zones containing clay
or chemically
disintegrated rock
(depth of excavation
>50 m)
D. Multiple shear zones 7.5
in competent rock
(clay free), loose
surrounding
E. Single shear zones in 5
rock (any depth)
competent rock (clay
free, depth of
excavation <50 m)
F. Single shear zones in 2.5
competent rock (clay
free, depth of
excavation >50 m)
G. Loose open joints, 5
heavily jointed, or
“sugar cube” etc (any
depth)
(b) Competent rock rock
stress problems
Strength/stress
ratios
σc/σ1 σt/σ1
H. Low stress, near >200 >13 2.5 (ii) If stress field is strongly anisotropic:
surface when 5<σ1/σ3< 10, reduce σc and σt to
0.8σc and 0.8σt; when σ1/σ3>10, reduce
J. Medium stress 200– 13– 1 σc and σt to 0.6σc and 0.6σt. Where
10 0.66 σc=unconfined compressive strength,
K. High stress, very tight 10–5 0.66– 0.5– σt=tensile strength, σ1 and σ3=major and
structure (usually 0.33 2.0 minor principal stresses.
favorable to stability,
maybe unfavorable to
wall stability)
L. Mild rock burst 5–2.5 0.33– 5–
(massive rock) 0.16 10
M. Heavy rock burst <2.5 <0.16 10–
(massive rock) 20
(c) Swelling rock;
chemical swelling
activity depending on (iii) Few case records available where depth
presence of water of crown below surface is less than span
width. Suggest SRF increase from 2.5 to
P. Mild swelling rock 5–10 5 for such cases (see H)
pressure
R. Heavy swelling rock 10–15
pressure
Table 2.16 Classification of rock mass based on Q-

values (after Barton et al., 1974).
Group Q Classification
1 1000−400 Exceptionally good
400–100 Extremely good
100−40 Very good
40–10 Good
2 10–4 Fair
4–1 Poor
1–0.1 Very poor
3 0.1–0.01 Extremely poor
0.01–0.001 Exceptionally poor
RMR=A loge Q+B

(2.6)
where A is typically in the range 5–15, and B in the range 35–60.

Based on data from hard rock tunneling projects in several countries, Barton (1991)
proposed a correlation between Q and seismic P-wave velocity:
(2.7)
where Vp is P-wave velocity in m/s (see Chapter 5 about the measurement of Vp).
2.3.4 Geological strength index (GSI)

Hoek and Brown (1997) introduced the Geological Strength Index (GSI), both for hard
and weak rock masses. Experienced field engineers and geologists generally show a
liking for a simple, fast, yet reliable classification which is based on visual inspection of
geological conditions. Hoek and Brown (1997) proposed such a practical classification
for estimating GSI based on visual inspection alone (see Tables 2.17). In this
classification, there are four main qualitative classifications, adopted from Terzaghi’s
classification (Terzaghi, 1946):
(i) Blocky
(ii) Very blocky
(iii) Blocky/Disturbed
(iv) Disintegrated
Engineers and geologists have been familiar with it for over 50 years. Further,
discontinuities are classified into 5 surface conditions which are similar to joint
conditions in RMR as described earlier:
(i)Very good
(ii) Good
(iii) Fair
(iv) Poor
(v) Very poor
Now a block in the 4×5 matrix of Table 2.17 is picked up according to actual rock mass
classification and discontinuity surface condition, and the corresponding GSI value can
then be read from the table. According to Hoek and Brown (1997), a range of values of
GSI (or RMR) should be estimated in preference to a single value.
GSI can also be estimated from Bieniawski’s Rock Mass Rating (RMR) and Barton et
al.’s (1974) modified Tunneling Quality Index (Q) (Hoek & Brown, 1997). In using
Bieniawski’s 1989 Rock Mass Rating (see Part A of Table 2.12) to estimate the value of
GSI, the rock mass should be assumed to be completely dry and a rating of 15 assigned to
the groundwater value. Very favorable discontinuity orientations should be assumed and
the Adjustment for Discontinuity Orientation value set to zero. The minimum value
which can be obtained for the 1989 classifications is 23. The estimated RMR is used to
estimate the value of GSI as follows:
(i)
Table 2.17 Characterisation of rock masses on the
basis of interlocking and joint alteration (after Hoek
& Brown, 1997).
For RMR>23
GSI=RMR−5
(2.8)
For RMR<23, Bieniawski’s classification cannot be used to estimate GSI and the Q value
of Barton et al. (1974) should be used instead:
GSI=9logeQ+44
(2.9)
where Q is calculated from Equation (2.5) by setting a value of 1 for both Jw

(discontinuity water reduction factor) and SRF (stress reduction factor).
It should be noted that, in the 1976 version of the Rock Mass Rating system, a rating
of 10 is assigned to the groundwater value when the rock mass is assumed to be
completely dry. The minimum value which can be obtained for the 1976 classifications is
thus 18. Using the 1976 Rock Mass Rating system, GSI is equal to the RMR obtained by
assuming completely dry rock mass condition and very favorable discontinuity
orientations.
3
Characterization of discontinuities in rock
3.1 INTRODUCTION
The primary difference between structures in soil and those in rock is that rock masses
contain discontinuities. The analysis and design of any structure in rock require
information on discontinuities in one form or another (Goodman, 1976; Hoek & Bray,
1981; Brady & Brown, 1985; Wyllie, 1999). Blocks formed by sets of unfavorably
orientated discontinuities may fail by sliding or toppling, causing excessive movement or
failure of foundations.
The discontinuity properties that have the greatest influence at the design stage have
been listed by Piteau (1970, 1973) as follows:
1. orientation
2. size
3. frequency
4. surface geometry
5. genetic type, and
6. infill material
As the rock discontinuities cannot be directly examined in three dimensions (3D),
engineers must infer discontinuity characteristics from data sampled at exposed rock
faces (including both natural outcrops and excavation walls) and/or in boreholes (Priest,
1993). Taking measurements at exposed rock faces, either at or below the ground surface,
enables one to obtain data on orientation, spacing, trace lengths and number of traces.
Many statistical sampling and data processing methods have been adopted (Priest, 1993).
The most widely used of these methods are the scanline and window sampling
techniques. These techniques have been described and discussed by a number of authors
including Baecher and Lanney (1978), ISRM (1978), Priest and Hudson (1981), Einstein
and Baecher (1983), Kulatilake and Wu (1984a, b, c), Kulatilake (1993), Mauldon (1998)
and Zhang and Einstein (1998b, 2000a). One disadvantage of this approach is that the
exposed rock face is often remote from the zone of interest and may suffer from blasting
damage or degradation by weathering and be obscured by vegetation cover.
Borehole sampling, in most cases, provides the only viable exploratory tool that
directly reveals geologic evidence of subsurface site conditions. In normal-size borehole
sampling various techniques can be used for acquiring discontinuity data either from core
samples or through inspection of the borehole walls. However, since the borehole
diameter is small, such information is of limited use only. In some cases, e.g., when
drilling holes for installing drilled shafts, the hole is large and direct measurements can
be conducted to obtain more accurate and precise discontinuity information. For example,
trace length data can be obtained from the side and bottom of the drilled shaft holes.
Characterization of discontinuities in rock 37
The data sampled at exposed rock faces and/or in boreholes contain errors due to
sampling biases. Therefore, it is important to correct sampling biases when inferring 3D
discontinuity characteristics. In addition, principles of stereology need be used in order to
infer 3D discontinuity characteristics from the data sampled at exposed rock faces and/or
in boreholes.
3.2 TYPES OF DISCONTINUITIES
Discontinuities and their origins are well described in several textbooks on general,
structural and engineering geology. From an engineer’s point of view, the discussions by
Price (1966), Hills (1972), Blyth and de Freitas (1974), Hobbs (1976) and Priest (1993)
are particularly helpful. The following lists the major types of discontinuities and briefly
describes their key engineering properties.
(a) Faults
Faults are discontinuities on which identifiable shear displacement has taken place. They
may be recognized by the relative displacement of the rock on the opposite sides of the
fault plane. The sense of this displacement is often used to classify faults. Faults may be
pervasive features which traverse a large area or they may be of relatively limited local
extent on the scale of meters; they often occur in echelon or in groups. Fault thickness
may vary from meters in the case of major, regional structures to millimeters in the case
of local faults. This fault thickness may contain weak materials such as fault gouge
(clay), fault breccia (recemented), rock flour or angular fragments. The wall rock is
frequently slickensided and may be coated with minerals such as graphite and chlorite
which have low frictional strengths. The ground adjacent to the fault may be disturbed
and weakened by associated discontinuities such as drag folds or secondary faults. These
factors result in faults being zones of low shear strength on which slip may readily occur.
(b) Bedding planes

Bedding planes divide sedimentary rocks into beds or strata. They represent interruptions
in the course of deposition of the rock mass. Bedding planes are generally highly
persistent features, although sediments laid down rapidly from heavily laden wind or
water currents may contain cross or discordant bedding. Bedding planes may contain
parting material of different grain size from the sediments forming the rock mass, or may
have been partially healed by low-order metamorphism. In either of these two cases,
there would be some ‘cohesion’ between the beds; otherwise, shear resistance on bedding
planes would be purely frictional. Arising from the depositional process, there may be a
preferred orientation of particles in the rock, giving rise to planes of weakness parallel to
the bedding planes.
(c) Joints
Joints are the most common and generally the most geotechnically significant
discontinuities in rocks. Joints are breaks of geological origin along which there has been
no visible relative displacement. A group of parallel or sub-parallel joints is called a joint
set, and joint sets intersect to form a joint system. Joints may be open, filled or healed.
Discontinuities frequently form parallel to bedding planes, foliations or slaty cleavage,
and they may be termed bedding joints, foliation joints or cleavage joints. Sedimentary
rocks often contain two sets of joints approximately orthogonal to each other and to the
bedding planes. These joints sometimes end at bedding planes, but others, called master
joints, may cross several bedding planes.
(d) Cleavage
There are two broad types of rock cleavage: fracture cleavage and flow cleavage.
Fracture cleavage (also known as false cleavage and strain slip cleavage) is a term
describing incipient, cemented or welded parallel discontinuities that are independent of
any parallel alignment of minerals. Spencer (1969) lists six possible mechanisms for the
formation of fracture cleavage. In each mechanism, lithology and stress conditions are
assumed to have produced shearing, extension or compression, giving rise to numerous
closely-spaced discontinuities separated by thin slivers of intact rock. Fracture cleavage is
generally associated with other structural features such as faults, folds and kink bands.
Flow cleavage, which can occur as slaty cleavage or schistosity, is dependent upon the
recrystallization and parallel allignment of platy minerals such as mica, producing inter-
leaving or foliation structure. It is generally accepted that flow cleavage is produced by
high temperatures and/or pressures associated with metamorphism in fine-grained rocks.
Although cleavage is usually clearly visible in slates, phyllites and schists, most
cleavage planes possess significant tensile strength and do not, therefore, contribute to the
discontinuity network. Cleavage can, however, create significant anisotropy in the
deformabilty and strength of such rocks. Geological processes, such as folding and
faulting, subsequent to the formation of the cleavage can exploit these planes of weakness
and generate discontinuities along a proportion of the better developed cleavage planes.
The decision as to whether a particular cleavage plane is a discontinuity presents one of
the most challenging problems to those undertaking discontinuity surveys in cleaved
rocks.
3.3 IMPORTANT PROPERTIES OF DISCONTINUITIES
This section lists and discusses the most important properties of discontinuities that
influence the engineering behavior. A fuller discussion of these properties can be found
in the ISRM publication Suggested methods for the quantitative description of
discontinuities in rock masses (ISRM, 1978).
Fig. 3.1 Definition of dip orientation

(α) and dip (β).
(a) Orientation
Orientation, or the attitude of a discontinuity in space, is described by the dip of the line
of maximum declination on the discontinuity surface measured from the horizontal, and
the dip direction or azimuth of this line, measured clockwise from true north (see Fig.
3.1). Some geologists record the strike of the discontinuity rather than the dip direction.
For rock mechanics purposes, it is usual to quote orientation data in the form of dip
direction (three digits)/dip (two digits) such as 035°/75° and 290°/30°. The orientation of
discontinuities relative to an engineering structure largely controls the possibility of
unstable conditions or excessive deformations developing. The importance of orientation
increases when other conditions for deformation are present, such as low shear strength
and a sufficient number of discontinuities for slip to occur. The mutual orientation of
discontinuities will determine the shape of the individual blocks, beds or mosaics
comprising the rock mass. The procedures for presenting and analyzing orientation
measurements using the stereonet are discussed in detail in Section 3.4.
(b) Spacing and frequency

Spacing is the perpendicular distance between adjacent discontinuities, and is usually
expressed as the mean spacing of a particular set of discontinuities. Frequency (i.e. the
number per unit distance) is the reciprocal of spacing (i.e. the mean spacing). The spacing
of discontinuities determines the sizes of the blocks making up the rock mass. The
mechanism of deformation and failure of rock masses can vary with the discontinuity
spacing. As in the case of orientation, the importance of spacing increases when other
conditions for deformation are present, such as low shear strength and a sufficient
number of discontinuities for slip to occur.
Like all other characteristics of a given rock mass, discontinuity spacings will not have
uniquely defined values but, rather, will take a range of values, possibly according to
some form of statistical distribution. The probabilistic analysis of discontinuity spacings
will be discussed in detail in Section 3.5.
Discontinuity spacing is a factor used in many rock mass classification schemes. Table
3.1 gives the terminology used by ISRM (1978).
Table 3.1 Classification of discontinuity spacing.
Description Spacing (mm)
Extremely close spacing <20
Very close spacing 20–60
Close spacing 60–200
Moderate spacing 200–600
Wide spacing 600–2000
Very wide spacing 2000–6000
Extremely wide spacing >6000
(c) Persistence and size

Persistence is a term used to describe the areal extent or size of a discontinuity within a
plane. It can be crudely quantified by observing discontinuity trace lengths on exposed
rock faces. It is one of the most important rock mass parameters, but one of the most
difficult to determine. The discontinuities of one particular set are often more continuous
than those of the other sets. The minor sets tend to terminate against the primary features,
or they may terminate in solid rock. The sets of discontinuities can be distinguished by
terms of persistent, sub-persistent and non-persistent respectively. Figure 3.2 shows a set
of simple plane sketches and block diagrams used to help indicate the persistence of
various sets of discontinuities in a rock mass. Clearly, the persistence of discontinuities
has a major influence on the shear strength developed in the plane of the discontinuity.
ISRM (1978) uses the most common or modal trace lengths of each set of
discontinuities measured on exposed rock faces to classify persistence according to Table
3.2.
Fig. 3.2 Simple sketches and block

diagrams indicating the persistence of
various sets of discontinuities (after
ISRM, 1978).
Table 3.2 Classification of discontinuity
persistence.
Description Modal trace length (m)
Very low persistence <1
Low persistence 1–3
Medium persistence 3–10
High persistence 10–20
Very high persistence >20
Persistence ratio PR is often used to describe the persistence of discontinuities. In the

literature, discontinuity persistence ratio PR is usually defined as
(3.1)
in which S is a region on the discontinuity plane with area As and is the area of the ith
discontinuity in S (see Fig. 3.3). The summation in Equation (3.1) is over all
discontinuities in S. Equivalently, discontinuity persistence ratio PR can be expressed as a
limit length ratio along a given line on a discontinuity plane. In this case,
(3.2)
in which LS is the length of a straight line segment S and is the length of the ith
discontinuity segment in S (see Fig. 3.4). For a finite sampling length LS, PR can be
simply estimated by (Fig. 3.5)
Fig. 3.3 Definition of PR as area ratio.
(3.3)
where ΣDL is the sum of the length of all discontinuities; and ΣRBL is the sum of the
length of all rock bridges.
Fig. 3.4 Definition of PR as length

ratio.
Fig. 3.5 Estimation of PR for a finite

sampling length.
Fig. 3.6 Failure of “low-angle”

transitions through intact rock: (a)
Tensile fracture; and (b) Secondary
shear fracture (after Zhang, 1999).
The above definition of discontinuity persistence ratio PR considers only the
discontinuities in the same plane. However, according to Einstein et al. (1983), when two
discontinuities are at a low-angle transition (β<θt, see Fig. 3.6), the rock bridge may fail
by the same mechanism as the in-plane rock bridge (see Fig. 3.7), where θt is the angle of
the tension cracks which can be obtained from Mohr’s circle [see Fig. 3.7(a)]. For both
the in-plane (Fig. 3.7) and the low-angle out-of-plane (Fig. 3.6) transitions, the intact-
rock resistance R can be calculated by
R=τad
(3.4)
where d is the “in-plane length” of the rock bridge; and τa is the peak shear stress
mobilized in the direction of discontinuities which can be obtained by
(3.5)
where σt is the tensile strength of the intact rock; and σa is the effective normal stress on
the discontinuity plane.
Fig. 3.7 In-plane failure of intact rock:

(a) Tensile fracture and corresponding
Mohr’s circle; and (b) Secondary shear
fracture (after Zhang, 1999).
Zhang (1999) proposed a definition of discontinuity persistence ratio PR that considers
both in-plane and low-angle-transition discontinuities:
(3.6)
in which LS is the total sampling length along the direction of the discontinuity traces,
DLi is the length of the ith in-plane discontinuities and DLl is the length of the lth low-
angle-transition discontinuities (see Fig. 3.8). For a finite sampling length, PR can be
simply approximated by
(3.7)
Fig. 3.8 Definition of PR considering

both in-plane and low-angle-transition
discontinuities (after Zhang, 1999).
where m and n are the numbers respectively of the in-plane and low-angle-transition (β
<θt) discontinuities within the sampling length LS (see Fig. 3.8).
(d) Shape
The planar shape of discontinuities has a profound effect on the connectivity of
discontinuities and on rock mass properties (Dershowitz et al., 1993; Petit et al., 1994).
However, since a rock mass is usually inaccessible in three dimensions, the real
discontinuity shape is rarely known. Information on discontinuity shape is limited and
often open to more than one interpretation (Warburton, 1980a; Wathugala, 1991).
Discontinuities can be classified into two categories: unrestricted and restricted.
Unrestricted discontinuities are blind and effectively isolated discontinuities whose
growth has not been perturbed by adjacent geological structures such as faults and free
surfaces. In general, the edge of unrestricted discontinuities is a closed convex curve. In

many cases, the growth of discontinuities is limited by adjacent preexisting
discontinuities and free surfaces. Such discontinuities are called restricted discontinuities.
One way to represent restricted discontinuities is to use polygons, some of the polygon
sides being those formed by intersections with the adjacent preexisting discontinuities
and free surfaces.
Due to the mathematical convenience, many investigators assume that discontinuities
are thin circular discs randomly located in space (Baecher et al., 1977; Warburton, 1980a;
Chan, 1986; Villaescusa & Brown, 1990; Kulatilake, 1993). With circular discontinuities,
the trace patterns in differently oriented sampling planes will be the same. In practice,
however, the trace patterns may vary with the orientation of sampling planes (Warburton,
1980b). Therefore, Warburton (1980b) assumed that discontinuities in a set are
parallelograms of various sizes. Dershowitz et al. (1993) used polygons to represent
discontinuities in the FracMan discrete fracture code. The polygons are formed by
inscribing a polygon in an ellipse. Ivanova (1998) and Meyer (1999) also used polygons
to represent discontinuities in their discrete fracture code GeoFrac. It is noted that
polygons can be used to effectively represent elliptical discontinuities when the number
of polygon sides is large (say >10) (Dershowitz et al., 1993). Zhang et al. (2002) assumed
that discontinuities are elliptical and derived a general stereological relationship between
trace length distributions and discontinuity size (expressed by the major axis length of the
ellipse) distributions.
Many researchers infer the discontinuity shape from the study of trace lengths in both
the strike and dip directions. Based on the fact that the average strike length of a
discontinuity set is approximately equal to its average dip length, Robertson (1970) and
Barton (1977) assumed that discontinuities are equidimensional (circular). However, the
average strike length of a discontinuity set being the same as its average dip length does
not necessarily mean that the discontinuities of such a set are equidimensional; instead,
there exist the following three possibilities (Zhang et al., 2002):
1. The discontinuities are indeed equidimensional [see Fig. 3.9(a)].
2. The discontinuities are non-equidimensional such as elliptical or rectangular with long
axes in a single (or deterministic) orientation. However, the discontinuities are
oriented such that the strike length is approximately equal to the dip length [see Fig.
3.9(b)].
3. The discontinuities are non-equidimensional such as elliptical or rectangular with long
axes randomly oriented. The random discontinuity orientation distribution makes the
average strike length approximately equal to the average dip length [see Fig. 3.9(c)].
Therefore, the conclusion that discontinuities are equidimensional (circular) drawn from
the fact that the average strike length of a discontinuity set is about equal to its average
dip length is questionable. Investigators assume circular discontinuity shape possibly
because of mathematical convenience.
Einstein et al. (1979) measured trace lengths of two sets of discontinuities on both the
horizontal and vertical surfaces of excavations and found that discontinuities are non-
equidimensional. Petit et al. (1994) presented results of a field study to determine the
shape of discontinuities in sedimentary rocks. Pelites with isolated sandstone layers in the
red Permian sandstones of the Lodeve Basin were studied. The exposed discontinuities
(i.e., one of the discontinuity walls had been removed by erosion) appear as rough
ellipses with a shape ratio L/H of about 2.0, where L and H are respectively the largest
horizontal and vertical dimensions. For non-exposed discontinuities, the distributions of
the dimensions of the horizontal and vertical traces were measured. The ratio of the mean
L to the mean H of such traces is 1.9, which is very close to the L/H ratio of the observed
individual discontinuity planes.
Fig. 3.9 Three possible cases for which

the average strike length is about equal
to the average dip length: (a)
Discontinuities are equidimensional
(circular); (b) Discontinuities are non-
equidimensional (elliptical), with long
axes in a single orientation. The
discontinuities are oriented so that the

average strike length is about equal to
the average dip length; and (c)
Discontinuities are non-
equidimensional (elliptical), with long
axes randomly orientated so that the
strike length is about equal to the dip
length (after Zhang et al., 2002).
(e) Roughness
Roughness is a measure of the inherent surface unevenness and waveness of the
discontinuity relative to its mean plane. The wall roughness of a discontinuity has an
important influence on its shear strength, especially in the case of undisplaced and
interlocked features such as unfilled joints. The importance of roughness declines with
increasing aperture, filling thickness or previous shear displacement. The important
influence of roughness on discontinuity shear strength is discussed in detail in Section
4.2.
When the properties of discontinuities are being recorded from observations made on
either borings cores or exposed faces, it is usual to distinguish between small-scale
surface irregularity or unevenness and large-scale undulations or waveness of the surface
(see Fig. 3.10). Each of these types of roughness may be quantified on an arbitrary scale
of, say, one to five. Descriptive terms may also be used particularly in the preliminary
stages of mapping. For example, ISRM (1978) suggests that the terms listed in Table 3.3
and illustrated in Figure 3.11 may be used to describe roughness on two scales—the
small scale (several centimeters) and the intermediate scale (several meters). Large-scale
waveness may be superimposed on such small- and intermediate-scale roughness.
Fig. 3.10 Different scales of

discontinuity roughness are sampled
by different scales of test. Waveness
can be characterized by the angle i
(after ISRM, 1978).
Table 3.3 Classification of discontinuity roughness.
Class Description
I Rough or irregular, stepped
II Smooth, stepped
III Slickensided, stepped
IV Rough or irregular, undulating
V Smooth, undulating
VI Slickensided, undulating
VII Rough or irregular, planar
VIII Smooth, planar
IX Slickensided, planar
Fig. 3.11 Typical roughness profiles

and suggested nomenclature. The
length of each profile is in the range of
1 to 10 meters. The vertical and
horizontal scales are equal (after
ISRM, 1978).
(f) Aperture
Aperture is the perpendicular distance separating the adjacent rock walls of an open
discontinuity in which the intervening space is filled with air or water. Aperture is
thereby distinguished from the width of a filled discontinuity (see Fig. 3.12). Large
apertures can result from shear displacement of discontinuities having appreciable
roughness, from outwash of filling materials (e.g. clay), from tensile opening, and/or
from solution. In most subsurface rock masses, apertures are small, probably less than
half a millimeter. Table 3.4 lists terms describing aperture dimensions suggested by
ISRM (1978). Clearly, aperture and its areal variation will have an influence on the shear
strength of discontinuities.
(g) Filling
Filling is a term used to describe material separating the adjacent rock walls of
discontinuities, such as calcite, chlorite, clay, silt, fault gourge, breccia, quartz and pyrite.
The perpendicular distance between the adjacent rock walls is termed the width of the
filled discontinuity, as opposed to the aperture of a gapped or open discontinuity. Filling
materials have a major influence on the shear strength of discontinuities. With the
exception of discontinuities filled with strong vein materials (calcite, quartz, pyrite),
filled discontinuities generally have lower shear strengths than comparable clean, closed
discontinuities. The behavior of filled discontinuities depends on many factors of which
the following are probably the most important:
Table 3.4 Classification of discontinuity aperture
Description Aperture (mm)
“Closed” features Very tight <0.1
Tight 0.1–0.25
Partly open 0.25–0.5
“Gapped” features Open 0.5–2.5
Moderately wide 2.5–10
Wide >10
“Open” features Very wide 10–100
Extremely wide 100–1000
Cavernous >1000
(1) Mineralogy of filling material

(2) Grading or particle size
(3) Over-consolidation ratio
(4) Water content and permeability
(5) Previous shear displacement
(6) Wall roughness
(7) Width
(8) Fracturing, crushing or chemical alteration of wall rock
Fig. 3.12 Suggested definitions of the

aperture of open discontinuities and
the width of filled discontinuities (after
ISRM, 1978).
If the filling materials are likely to influence the performance of foundations, samples of
the filling materials (undisturbed if possible) should be collected, or an in situ test may be
carried out.
3.4 STEREOGRAPHIC PROJECTION
Stereographic projection is a procedure for mapping data located on the surface of a

sphere on to a horizontal plane, and can be used for analysis of the orientation of planes,
lines and forces (Donn & Shimmer, 1958; Phillips, 1971; Goodman, 1976; Hoek & Bray,
1981). There are several types of stereographic projections, but the one most suitable for
geological applications is the equal area net, or Lambert projection, which is also used by
geographers to represent the spherical shape of the Earth on a flat surface. In structural
geology, a point or a line on the sphere representing the dip and dip direction of a
discontinuity can be projected on to a horizontal surface. In this way an analysis of three-
dimensional data can be carried out in two dimensions.
3.4.1 Hemispherical projection of a plane

Consider a reference sphere which is oriented in space, usually with respect to true north.
When a plane (discontinuity) passes through the center of the reference sphere, the
intersection between the plane and the sphere surface is a circle which is called the great
circle. A line normal to the plane and passing through the center of the sphere intersects
the sphere at two diametrically opposite points called the poles of the plane. Because the
great circle and the pole representing the plane appear on both the upper and lower halves
of the sphere, only one hemisphere need be used to plot and manipulate structural data. In
rock mechanics and rock engineering, the lower hemisphere projection is almost always
used.
Figure 3.13 shows the lower hemisphere projection of a great circle and its pole onto the
horizontal plane passing through the center of the sphere. This is known as the equal-
angle or Wulff projection. In this projection, any circle on the reference hemisphere
projects as a circle on the plane of the projection. This is not the case for an alternative
projection known as equal-area or Lambert projection. The latter is better suited than the
equal-angle projection for use in the analysis of discontinuity orientation. The equal-area
projection for any point on the surface of the reference sphere is accomplished by
drawing an arc about the lower end of the vertical axis of the sphere from the point to the
horizontal base plane (see Fig. 3.14).
Fig. 3.13 (a) Stereographic projection

of a great circle and its pole from the
lower reference hemisphere onto the

horizontal plane passing through the
center of the reference sphere; and (b)
Plan view of the equal-angle projection
of the great circle and pole of a plane
230/50.
Stereographic projections of the great circle and pole of a discontinuity can be prepared
by hand by plotting the data on standard sheets (stereonets) with lines representing dip
and dip direction values. For the detailed steps, the reader can refer to books on structural
geology and rock mechanics such as those by Phillips (1971), Goodman (1976) and Hoek
and Bray (1981). Alternatively, there are computer programs such as the one by Mahtab
et al. (1972) available that will not only plot great circles and poles, but will also obtain
orientation frequency contour plots.
It should be noted that the projection technique only examines the orientation of
discontinuities and there is no information on their position in space. That is, it is
assumed that all the planes pass through the center of the reference sphere. If the
stereographic projection identifies a plane on which the drilled shaft foundation could
slide, its location on the geological map will have to be examined to determine if it
intersects the drilled shaft foundation.
3.4.2 Plotting and analysis of discontinuity orientation data

An elementary use of the stereographic projection is the plotting and analysis of field
measurements of discontinuity orientation data. In plotting field measurements of dip and
dip direction, it is convenient to work with poles rather than great circles since the poles
can be plotted directly on a polar stereonet such as that shown in Figure 3.15.
Fig. 3.14 (a) Vertical section through

reference sphere showing lower
hemisphere, equal-area projection of a
great circle and its pole; and (b) Plan
view of the equal-area projection of the
great circle and pole of a plane 230/50.
Fig. 3.15 Polar stereographic net for

plotting poles of discontinuity planes.
Suppose that a plane has dip direction and dip values 030/40, the pole is located on the
stereonet by using the dip direction value of 30 given in bold and then measuring the dip
value of 40 from the center of the net along the radial line.
Figure 3.16 shows such a plot of the poles of 351 individual discontinuities whose
orientations were measured at a particular field site (Hoek & Brown, 1980). Different
symbols have been used for three different types of discontinuities—bedding planes,
joints and a fault. The fault has a dip direction of 307° and a dip of 56°. Contours of pole
concentrations may be drawn for the bedding planes and joints to give an indication of
the preferred orientations of the various discontinuity sets present. The essential tool
required for pole contouring is a counting net which divides the surface of the reference
hemisphere into a number of equal areas. Figure 3.17 shows a counting net containing
100 equal areas for use with the polar stereonet shown in Figure 3.15. The most
convenient way of using the counting net is to prepare a transparent overlay of it and to
center this overlay on the pole plot by means of a pin through the center of the net. A
piece of tracing paper is mounted on top of the overlay, pierced by the center pin but
fixed by a piece of adhesive tape so that it cannot rotate with respect to the pole plot.
Keeping the counting net in a fixed position, the number of poles falling within each
counting cell can be counted and noted in pencil on the tracing paper at the center of each
cell. The counting net is then rotated to center the densest pole
Fig. 3.16 Plot of poles of 351

discontinuities (after Hoek & Brown,
1980).
Fig. 3.17 Counting net used in

conjunction with the polar
stereographic net shown in Figure
3.15.
Fig. 3.18 Contours of pole

concentrations for the data plotted in
Figure 3.15 (after Hoek & Brown,
1980).
concentrations in counting cells and the maximum percentage pole concentrations are
determined. By further small rotations of the counting net, the contours of decreasing
percentage which surround the maximum pole concentrations can be established. Figure
3.18 shows the contours of pole concentrations determined in this way from the data
shown in Figure 3.16. The central orientations of the two major two joint sets are 347/22
and 352/83, and that of the bedding planes is 232/81.
Computer programs such as the one by Mahtab et al. (1972) are also available for
plotting and contouring discontinuity orientation data. For a large number of
discontinuities, it will be more convenient to use computer programs to plot and contour
orientation data.
The assignment of poles into discontinuity sets is usually achieved by a combination
of contouring, visual examination of the stereonet and a knowledge of geological
conditions at the site. However, in many cases visual clustering is very difficult due to the
overlap of clusters. A number of algorithms which are based on statistical or fuzzy-set
approaches are available for numerically clustering orientation data (Einstein et al., 1979;
Miller, 1983; Mahtab & Yegulalp, 1984; Harrison, 1992; Kulatilake, 1993).
3.5 STATISTICAL ANALYSIS OF DISCONTINUITY

PROPERTIES
Since discontinuity characteristics are inherently statistical, statistical techniques are

widely used in the data reduction, presentation and analysis of discontinuity properties.
3.5.1 Discontinuity orientation

As seen in Figure 3.16, there is scatter of the poles of discontinuities when they are
plotted on the stereonet. The mean orientation of a number of discontinuities can be
calculated from the direction cosines as follows. The sampling bias on orientation can
also be considered.
The pole of a discontinuity in three-dimensional space can be represented by a unit
vector (ux, uy, uz) associated with the direction cosines as shown in Figure 3.19:
ux=cos αn cos βn, uy=sin αn cos βn, uz=sin βn
(3.8)
where αn and βn are respectively the trend and plunge of the pole, which can be obtained
by
(3.9a)
(3.9b)
The parameter Q is an angle, in degrees that ensures that αn lies in the correct quadrant
and in the range of 0 to 360° (see Table 3.5).
Fig. 3.19 Pole of a discontinuity

represented by a unit vector.
Tabte 3.5 The quadrant parameter Q in Equation
(3.9a).
ux uy Q
≥0 ≥0 0
<0 ≥0 180˚
<0 <0 180°
≥0 <0 360˚
The dip direction and dip angle α/β of a discontinuity are related to the trend and plunge
αn/βn of its normal by the following expressions:
αn=α+180° (for α≤180°)
αn=α−180° (for α≥180°) (3.10a)
βn=90°−β
(3.10b)
The mean orientation of a set of discontinuities intersecting a sampling line of

trend/plunge αs/βs can be obtained using the procedure outlined below. This procedure
corrects for orientation sampling bias through the introduction of weighted direction
cosines.
1. For discontinuity i, calculate the angle δi between its normal and the sampling line:
cos δi=|uxiuxs+uyiuys+uziuzs|
(3.11)
where (uxi, uyi, uzi) and (uxs, uys, uzs) are the direction cosines respectively of
the normal to discontinuity i and the sampling line.
2. For discontinuity i, calculate the weighting factor wi based on the angle δi obtained
in step 1:
(3.12)
3. After the weighting factor for each discontinuity is obtained, calculate the total
weighted sample size Nw for a sample of size N by
(3.13)
4. Calculate the normalized weighting factor wni for each discontinuity by
(3.14)
5. Calculate the corrected direction cosines (nxi, nyi, nzi) for the normal of each
discontinuity by
(nxi, nyi, nzi)=wni(uxi, uyi, uzi)
(3.15)
6. Calculate the resultant vector (rx, ry, rz) of the corrected normal vectors (nxi, nyi, nzi),
i=1 to N:
(3.16)
7. The mean orientation of the N discontinuities is the orientation of the resultant

vector whose trend and plunge can be found by replacing ux, uy and uz by rx, ry and
rz in Equation (3.9).
Several probability distributions have been suggested in the literature to represent the
discontinuity orientations (Shanley & Mahtab, 1976; Zanbak, 1977; Einstein et al., 1979;
Kulatilake, 1985a, 1986). The distributions are the hemispherical uniform, hemispherical
normal or Fisher, bivariate Fisher, Bingham, bivariate normal and bivariate lognormal.
The best means to check if a certain distribution is applicable to represent the orientation
of a discontinuity set is to perform goodness-of-fit tests. Shanley and Mahtab (1976) and
Kulatilake (1985a, 1986) have presented χ2 goodness-of-fit tests respectively for

Bingham, hemispherical normal and bivariate normal distributions. Einstein et al. (1979)
have tried all the aforementioned distributions to represent the statistical distributions for
22 data sets. They have reported that they could not find a probability distribution which
satisfied χ2 test at 5 percent significance level for 18 of these data sets. This shows clearly
the inadequacy of the currently available analytical distributions in representing the
discontinuity orientations. In the case that no analytical distribution can represent the
discontinuity orientation data, empirical distributions can be used.
3.5.2 Discontinuity spacing and frequency

Discontinuity spacings usually take a range of values, possibly according to some form of
statistical distribution. Priest and Hudson (1976) made measurements on a number of
sedimentary rock masses in the United Kingdom and found that, in each case, the
discontinuity spacing histogram gave a probability density distribution that could be
approximated by the negative exponential distribution. The same conclusion has been
reached by others, notably Wallis and King (1980) working on a Precambrian porphyritic
granite, and Baecher (1983) working on a variety of igneous, sedimentary and
metamorphic rocks. Thus the frequency f(s) of a given discontinuity spacing value s is
given by the function
f(s)=λe−λs
(3.17)
where is the mean discontinuity frequency of a large discontinuity population

and is the mean spacing.
Figure 3.20 shows the discontinuity spacing histogram and the corresponding negative
exponential distribution calculated from Equation (3.17) for the Lower Chalk, Chinnor,
Oxfordshire, UK (Priest & Hudson, 1976). The use of frequency distributions such as that
given by Equation (3.17) permits statistical calculations to be made of such factors as
possible block sizes and the likelihood that certain types of intersection will occur.
Priest and Hudson (1976) found that an estimate of RQD (see Chapter 2 about the
definition of RQD) could be obtained from discontinuity spacing measurements made
Fig. 3.20 Discontinuity spacing

histogram for all scanlines in the first
85 m of tunnel, Chinnor UK (after
Priest & Hudson, 1976).
on core or exposed rock face using the equation
RQD=100e−0.1λ(0.1λ+1)
(3.18)
For values of λ in the range 6 to 16 m−1, a good approximation to measured RQD values
was found to be given by the linear relation
RQD=−3.68λ+110.4
(3.19)
Figure 3.21 shows the relations obtained by Priest and Hudson (1976) between measured
values of RQD and λ, and the values calculated using Equations (3.18) and (3.19). Please
note the similarity of Equation (3.19) and Equation (2.2).
3.5.3 Discontinuity trace length

In sampling for trace lengths, errors can occur due to the following biases (Baecher &
Lanney, 1978; Einstein et al., 1979; Priest & Hudson, 1981; Kulatilake & Wu, 1984c;
Mauldon, 1998; Zhang & Einstein, 1998b, 2000a):
(1) Orientation bias: the probability of a discontinuity appearing at an exposed rock face
depends on the relative orientation between the rock face and the discontinuity.
(2) Size bias: large discontinuities are more likely to be sampled than small
Fig. 3.21 Relationship between RQD

and mean discontinuity frequency
(after Priest & Hudson, 1976).
discontinuities. This bias affects the results in two ways: (a) a larger discontinuity
is more likely to appear at an exposed rock face than a smaller one; and (b) a
longer trace is more likely to appear in a sampling area than a shorter one.
(3) Truncation bias: Very small trace lengths are difficult or sometimes impossible to
measure. Therefore, trace lengths below some known cutoff length are not recorded.
(4) Censoring bias: Long discontinuity traces may extend beyond the visible exposure so
that one end or both ends of the discontinuity traces can not be seen.
In inferring the trace length distribution on an infinite surface from the measured trace
lengths on a finite size area on this surface, biases (2b), (3) and (4) should be considered.
Biases (1) and (2a) are of interest only in three-dimensional simulations of
discontinuities, i.e., when inferring discontinuity size distributions as discussed in Section
3.5.4.
(a) Probability distribution of measured trace lengths

Many investigators have looked into the distribution of trace lengths (Table 3.6). Apart
from Baecher et al. (1977), Cruden (1977), Einstein et al. (1979) and Kulatilake (1993)
others have based their argument on inspection rather than on goodness-of-fit tests. It
seems that only Baecher et al. (1977), Einstein et al. (1979) and Kulatilake (1993) have
tried more than one distribution to find the best distribution to represent trace length data.
To find the suitable distribution for the measured trace lengths of each discontinuity
set, the distribution forms in Table 3.6 can be checked by using χ2 and Kolmogorov-
Smirnov goodness-of-fit tests.
(b) Corrected mean trace length

In inferring the corrected mean trace length (i.e., the mean trace length on an infinite
surface), from the measured trace lengths on a finite exposure, biases (2b), (3) and (4)
should be considered. Truncation bias (3) can be corrected using the method of
Table 3.6 Distribution forms of trace length.
Investigator Distribution
Robertson (1970) Exponential
McMahon (1974) Lognormal
Bridges (1975) Lognormal
Call et al. (1976) Exponential
Barton (1977) Lognormal
Cruden (1977) Exponential
Baecher et al (1977) Lognormal
Einstein et al. (1979) Lognormal
Priest and Hudson (1981) Exponential
Kulatilake (1993) Exponential and Gamma (Gamma better)
Warburton (1980a). Decreasing the truncation level in discontinuity surveys can reduce
effects of truncation bias on trace length estimates. It is practically feasible to observe
and measure trace lengths as low as 10 mm both in the field and from photographs (Priest
& Hudson, 1981). Truncation at this level will have only a small effect on the data,
particularly if the mean trace length is in the order of meters (Priest & Hudson, 1981;
Einstein & Baecher, 1983). Therefore, the effect of truncation bias on trace length
estimates can be ignored. However, biases (2b) and (4) are important (Kulatilake & Wu,
1984c) and need be considered.
Pahl (1981) suggested a technique to estimate the mean trace length on an infinite
surface produced by a discontinuity set whose orientation has a single value, i.e., all
discontinuities in the set have the same orientation. His technique is based on the
categorization of randomly located discontinuities that intersect a vertical, rectangular
planar rock face window of height h and width w, and whose traces make an angle with
the vertical, as shown in Figure 3.22. Discontinuities intersecting the sampling window
can be divided into three classes: (1) discontinuities with both ends censored, (2)
discontinuities with one end censored and one end observable, and (3) discontinuities
with both ends observable. If the numbers of traces in each of the above three types are
N0, N1 and N2 respectively, the total number of traces, N, will be
N=N0+N1+N2
(3.20)
Pahl (1981) has derived the following expression for mean trace length µ
(3.21)
Although the approach in Equation (3.21) is both rigorous and easy to implement, it
Fig. 3.22 Discontinuities intersecting a

vertical rock face (after Pahl, 1981).
does rely on the discontinuities being grouped into a parallel or nearly parallel set.
Kulatilake and Wu (1984c) extended Pahl’s technique to discontinuities whose
orientation is described by a probabilistic distribution. A major difficulty in applying the
extended technique is to determine the probabilistic distribution function of the
orientation of discontinuities.
Using different methods, Mauldon (1998) and Zhang and Einstein (1998b)
independently derived the following expression for estimating the mean trace length on
an infinite surface, from the observed trace data in a finite circular window (see Fig.
3.23):
(3.22)
where c is the radius of the circular sampling window. The major advantage of the
method of Mauldon (1998) and Zhang and Einstein (1998b) over the methods of Pahl
(1981) and Kulatilake and Wu (1984c) is that it does not need sampling data about the
orientation of discontinuities, i.e., the method of Mauldon (1998) and Zhang and Einstein
(1998b) is applicable to traces with arbitrary orientation distributions. Therefore, the
method of Mauldon (1998) and Zhang and Einstein (1998b) can be used to estimate the
mean trace length of more than one set of discontinuities. The orientation distribution-
free nature of this method comes from the symmetric properties of the circular sampling
windows.
Trace length measurements are not needed when using Equations (3.21) and (3.22). In
the derivation of Equations (3.21) and (3.22), discontinuity trace length l can be
anywhere between zero and infinity. Hence, µ obtained by Equations (3.21) and (3.22)
does not contain errors due to biases (2b) and (4) as described before.
µ in Equations (3.21) and (3.22) is the population (thus correct or true) mean trace
length, with N, N0 and N2 being respectively the expected total number of traces
intersecting the window, the expected number of traces with both ends censored and the
expected number of traces with both ends observable. In practice, the exact values of N,
N0 and N2 are not known and thus µ has to be estimated using sampled data. From
sampling in one rectangular or circular window, what we get is only one sample of N, N0
and N2 and from this sample only a point estimate of µ can be obtained. For example, for
a sample of traces intersecting the sampling window, if and are respectively

the numbers of discontinuities that appear on the window with both ends censored and
both ends observable, the mean trace length of the sample, , can be obtained by
(3.23)
(3.24)
Fig. 3.23 Discontinuities intersecting a

circular sampling window (after Zhang
& Einstein, 1998b).
In other words, the of several samples can be used to evaluate µ.

When applying Equation (3.23) or (3.24), the following two special cases may occur:
(1) If , then . In this case, all the discontinuities intersecting the

sampling window have both ends censored. This implies that the area of the window
used for the discontinuity survey may be too small.
(2) If , then . In this case, all the discontinuities intersecting the

sampling window have both ends observable. According to Pahl (1981), this results is
due to violation of the assumption that the midpoints of traces are uniformly
distributed in the two dimensional space.
These two special cases can be addressed by increasing the sampling window size and/or
changing the sampling window position (Zhang, 1999). Another method to address these
two special cases is to use multiple windows of the same size but at different locations
and then use the total numbers from these windows to estimate (Zhang & Einstein,
1998b).
(c) Trace length distribution on an infinite surface

Two probability density functions (pdf) can be defined for trace lengths as follows:
(1) f(l)=pdf of trace lengths on an infinite surface.
(2) g(l)=pdf of measured trace lengths on a finite exposure subjected to sampling biases.
It is necessary to obtain f(l) from g(l), because 3D size distribution of discontinuities is
inferred from f(l). Zhang and Einstein (2000a) proposed the following procedure for
obtaining f(l):
(1) Use the corrected mean trace length µ as the mean value of f(l)
(2) Use the coefficient of variation (COV) value of g(l) as the COV of f(l)
(3) Find the distribution of g(l) as discussed before and assume that f(l) and g(l) have the
same distribution form
3.5.4 Discontinuity size

Zhang and Einstein (2000a) presented a method for inferring the discontinuity size
distribution from the corrected trace length distribution obtained from circular window
sampling as described in Section 3.5.3, based on the stereological relationship between
trace lengths and discontinuity diameter distributions for area sampling of discontinuities
(Warburton, 1980a):
(3.25)
where D is the diameter of discontinuities; l is the trace length of discontinuities; g(D) is

the probability density function of the diameter of discontinuities; f(l) is the probability
density function of the trace length of discontinuities; and µD is the mean of the diameter
of discontinuities. Villaescusa and Brown (1992) presented a similar method for inferring
the discontinuity size distribution from the corrected trace length distribution obtained
from straight scanline sampling. They used the following stereological relationship
between trace length and discontinuity diameter distributions for straight scanline
sampling of discontinuities (Warburton, 1980a):
(3.26)
where E(D2) is the mean of D2.

Zhang et al. (2002) derived a general stereological relationship between trace length
distributions and discontinuity size (expressed by the major axis length a of the ellipse)
distributions for area (or window) sampling, following the methodology of Warburton
(1980a, b):
(3.27)
where
(3.28)
in which k is the aspect ratio of the discontinuity, i.e., the length of the discontinuity
minor axis is a/k (see Fig. 3.24); β is the angle between the discontinuity major axis and
the trace line (note that β is measured in the discontinuity plane). Obviously, β will
change for different sampling planes. For a specific sampling plane, however, there will
be only one β value for a discontinuity set with a deterministic orientation.
When k=1 (i.e., the discontinuities are circular), M=1 and Equation (3.27) reduces to
Equation (3.25).
Based on Equation (3.27), Zhang et al. (2002) extended the method of Zhang and
Einstein (2000a) to elliptical discontinuities. Table 3.7 summarizes the expressions for
determining µa and σa from µl and σl, respectively for the lognormal, negative exponential
and Gamma distribution of discontinuity size a. Conversely, with known µa and σa, and
the distribution form of g(a) the mean µl and standard deviation σl of trace lengths can
also be obtained (see Table 3.8).
Consider a discontinuity set having a lognormal size distribution with µa=8.0 m and
σa=4.0 m (For other distribution forms, similar conclusions can be obtained). Figure 3.25
shows the variation of the mean trace length and the standard deviation of trace lengths
with β. Since β is the angle between the trace line and the discontinuity major axis, it is
related to the sampling plane orientation relative to the discontinuity. It can be seen that,
despite the considerable difference between the maximum and the minimum,
Fig. 3.24 Parameters used in the

definition of an elliptical discontinuity
(after Zhang et al., 2002).
respectively, of the mean trace length and the standard deviation of trace lengths, there
are extensive ranges of sampling plane orientations, reflected by β, over which both the
mean trace length and the standard deviation of trace lengths show little variation,
especially for large k values. The results in Figure 3.25 could well explain why Bridges
(1976), Einstein et al. (1979) and McMahon (Mostyn & Li, 1993) found different mean
trace lengths on differently oriented sampling planes, whereas Robertson (1970) and
Barton (1977) observed them to be approximately equal. In each of these papers or
reports, the number of differently oriented sampling planes was very limited and,
depending on the relative orientations of the sampling planes, the authors could observe
either approximately equal mean trace lengths or significantly different mean trace
lengths. For example, in Bridges (1976), Einstein et al. (1979) and McMahon (Mostyn &
Li, 1993), the strike and dip sampling planes might be respectively in the β=0°–20° (or
160°–180°) range and the β=40°–140° range, or vice versa. From Figure 3.25, this would
result in very different mean trace lengths. On the other hand, in Robertson (1970) and
Barton (1977), the strike and dip sampling planes might be both in the β=40°–140° range
(i.e., in the “flat” trace length part of Fig. 3.25) or respectively in some β ranges
approximately symmetrical about β=90°. It should be noted that the comments above are
assumptions because no information about the β values can be found in the original
papers or reports.
The implications of Figure 3.25 about field sampling are as follows:
If different sampling planes are used to collect trace (length) data, the
sampling planes should be oriented such that significantly different mean
trace lengths can be obtained from different planes. For example, if two
sampling planes are used, one should be oriented in the β=0°–20° (or 160°
–180°) range and the other in the β=60°–120° range.
Table 3.7 Expressions for deteimining µa and σa
from µl and σl.
Distribution µa (σa)2
form of
g(a)
Lognormal
Negative
exponential
Gamma
Table 3.8 Expressions for determining µl and σl

from µa and σa.
Distribution µl (σl)2
form of g(a)
Lognormal
Negative
exponential
Gamma
Fig. 3.25 Variation of mean trace

length and standard deviation (s.d.) of
trace length with β.
It is noted that, with the same µl and σl, one can have different µa and σa if the assumed
distribution form of g(a) is different. This means that the estimation of discontinuity size
distributions from the equations in Table 3.7 may not be robust. To overcome the
problem of uniqueness, a relationship between the ratio of the 4th and 1st moments of the
discontinuity size distribution and the 3rd moment of the trace length distribution is used
to check the suitability of the assumed discontinuity size distribution form:
(3.29)
For the three distribution forms of g(a) discussed above, Equation (3.29) can be rewritten
as:
(a) If g(a) is lognormally distributed with mean µa and standard deviation σa,
(3.30)
(b) If g(a) has a negative exponential distribution with mean µa,
(3.31)
(c) If g(a) has a Gamma distribution with mean µa and standard deviation σa,
(3.32)
The procedure for inferring the major axis orientation, aspect ratio k and size distribution
g(a) (probability density function of the major axis length) of elliptical discontinuities
from trace length sampling on different sampling windows is summarized as follows (the
reader can refer to Zhang et al., 2002 for details):
1. Sampling
(a) Trace length: Use two or more sampling windows at different orientations to
conduct trace (length) sampling. The sampling windows (planes) should be
oriented such that significantly different mean trace lengths can be obtained from
different windows.
(b) Orientation: Use exposed rock surface or borehole sampling so that the normal
orientation of each discontinuity set can be obtained.
2. Conduct trace length analysis to estimate the true trace length distribution f(l) on
different sampling windows: µl, σl and form of f(l).
3. Infer the major axis orientation, aspect ratio k and size distribution g(a) of
discontinuities from trace length sampling on different sampling windows:
(a) Assume a major axis orientation and compute the β (the angle between
discontinuity major axis and trace line) value for each sampling window.
(b) For the assumed major axis orientation, compute µa and σa from µl and σl of each
sampling window, by assuming aspect ratios k=1, 2, 4, 6, 8 and lognormal,
negative exponential and Gamma distribution forms of g(a). The results are then
used to draw the curves relating µa (and σa) to k, respectively, for the lognormal,
negative exponential and Gamma distribution forms of g(a).
(c) Repeat steps (a) and (b) until the curves relating µa (and σa) to k for different
sampling windows intersect in one point. The major axis orientation for this case is
the inferred actual major axis orientation. The k, µa and σa values at the intersection
points are the corresponding possible characteristics of the discontinuities.
(d) Find the best distribution form of g(a) by checking the equality of Equation (3.29).
The k, µa and σa values found in Step (c) and corresponding to the best distribution
form of g(a) are the inferred characteristics of the discontinuity size.
3.6 FRACTURE TENSOR FOR DESCRIBING DISCONTINUITY

GEOMETRY
Tensors have been used by several researchers to describe discontinuity geometry

including intensity and orientation. Kachanov (1980) introduced a tensor αij to quantify
the geometry of microcracks in rocks
(3.33)
where V is the volume of the rock mass considered; S(k) is the area of the kth
discontinuity; m(V) is the number of discontinuities in volume V; ui(k) and uj(k) (i, j=x, y, z)
are components of the unit normal vector of the kth discontinuity with respect to
orthogonal reference axes i and j (i, j=x, y, z) respectively (see Fig. 3.19 about the
definition of the normal direction of a discontinuity).
Oda (1982) also proposed a tensor Fij (called the crack tensor) for describing
discontinuity geometry
(3.34)
where r(k) is the radius of the kth discontinuity.

Kawamoto et al. (1988) regarded discontinuities as damages, and defined a tensor Ωij
called the damage tensor
(3.35)
where is a characteristic length for a given discontinuity system.

The tensors described above are non-dimensional due to some arbitrary operation
included in their definitions: in Equation (3.33) the area S(k) of a discontinuity is
multiplied by the square root of S(k); in Equation (3.34) the area S(k) of a discontinuity is
multiplied by its radius r(k); and in Equation (3.35) a characteristic length for a given
discontinuity system is included. Because of the arbitrary operation, the physical meaning
of the discontinuity intensity expressed by those definitions is not clear and thus a little
confusing (e.g., what is the physical meaning of [S(k)]3/2?).
Dershowitz and Herda (1992) and Mauldon (1994) have shown that P32, defined as the
mean area of discontinuities per unit volume of the rock mass, is the most useful measure
of discontinuity intensity. However, P32 does not include the effect of discontinuity
orientations. Zhang (1999) introduced the fracture tensor Fij, which is a combined
measure of discontinuity intensity and orientation, defined as follows:
(3.36)
Fij can be determined with the data obtained in the previous sections. Fracture tensor Fij
can also be written in matrix form as follows
(3.37)
Fij has three principal values F1, F2 and F3, which can be obtained by finding the
eigenvalues of Fij. The principal orientation of Fij can be obtained by finding the
eigenvectors corresponding to F1, F2 and F3.
The first invariant of Fij is just P32, i.e.,
(3.38)
In contrast to the tensors proposed by Kachanov (1980), Oda (1982) and Kawamoto et al.
(1988), the fracture tensor defined in Equation (3.36) has a clear physical meaning. It
represents the ratio of the total area of discontinuities and the volume of the rock mass
considered. The fracture tensor defined in Equation (3.36) keeps the advantage of P32,
i.e., P32 does not depend on the size of the sampled region as long as it is representative
of the discontinuity network.
4
Deformability and strength of rock
4.1 INTRODUCTION
Discontinuities have a profound effect on the deformability and strength of rock masses.
Characterization of a rock mass depends not only on the nature of the rock material, but
also on the discontinuities which are pervasive throughout almost all natural rock. The
presence of discontinuities has long been recognized as an important factor influencing
the mechanical behavior of rock masses. The existence of one or several discontinuity
sets in a rock mass creates anisotropy in its response to loading and unloading. Also,
compared to intact rock, jointed rock shows reduced shear strength along planes of
discontinuity, increased deformability parallel to those planes, and increased
deformability and negligible tensile strength in directions normal to those planes.
One of the most important concerns in designing foundations in rock is the
determination of deformation and strength properties of rock masses. For predicting the
ultimate load capacity of a foundation in rock, a strength model of the rock mass is
required. Alternatively, if predictions of the foundation movements caused by the applied
loading are required, a constitutive (or deformation) model must be selected. With few
exceptions, it is incorrect to ignore the presence of discontinuities when modeling rock
mass response to loading and unloading. Therefore, it is important to account for the
effect of discontinuities on the deformation and strength properties of rock masses.
Currently, there are two ways to account for the effect of discontinuities on the
deformation and strength properties of jointed rock masses: direct and indirect methods
(Amadei & Savage, 1993; Kulatilake, et al. 1992, 1993; Wang & Kulatilake, 1993).
Direct Methods
Direct methods include laboratory and in situ tests, To obtain realistic results of rock
mass deformability and strength, rock of different volumes having a number of different
known discontinuity configurations should be tested at relevant stress levels under
different stress paths. Such an experimental program is almost impossible to carry out in
the laboratory. With in situ tests, such an experimental program would be very difficult,
time-consuming and expensive. At the laboratory level, some researchers have performed
experiments on model material to study mechanical behavior of jointed rock. Results of
laboratory model studies (Brown, 1970a, b; John, 1970; Ladanyi & Archambault, 1970;
Einstein & Hirschfeld, 1973; Chappel, 1974; Singh et al., 2002) show that many different
failure modes are possible in jointed rock and that the internal distribution of stresses
within a jointed rock mass can be highly complex. Since laboratory tests on small scale
samples are often inadequate to predict the deformability and strength of rock masses, in
situ tests are necessary. There are many types of in situ tests, including uniaxial
compression, plate bearing, flat jack, pressure chamber, borehole jacking and dilatometer
Deformability and strength of rock 81
tests. Deformability properties may be estimated from such in situ test data, usually
assuming that some idealized model describes the rock behavior in the test configuration.
A few in situ tests have been carried out to study the effect of size on rock mass
compressive strength (Bieniawski, 1968; Pratt et al., 1972; Bieniawski & Van Heerden,
1975) and on rock mass modulus (Bieniawski, 1978). Heuze (1980) has reviewed the
previous work on scale effects on mass deformability and strength. The results of these
investigations clearly show the reduction of mass strength and modulus with size up to a
certain size at which changes become insignificant. It is important to note that these
relations are highly site-dependent, since the scale effect is primarily governed by the
discontinuity networks. Chapter 5 will discuss the in situ tests in more detail.
Indirect Methods
The indirect methods can be divided into the following three approaches:
1. The first indirect approach consists of empirically deducing the deformability and
strength properties of rock masses from those measured on intact rock samples in the
laboratory. Rock mass modulus and strength can be estimated in different ways. Deere
et al. (1967), Coon and Merritt (1970), Gardner (1987) and Zhang and Einstein
(2000b) presented correlations between rock quality designation (RQD) and modulus
ratio Em/Er, where Em and Er are respectively the rock mass deformation modulus and
the intact rock deformation modulus. Bieniawski (1978) and Serafim and Pereira
(1983) proposed relationships between the deformation modulus of rock masses and
the RMR ratings using the geomechanics classification system (Bieniawski, 1974).
Based on practical observations and back analysis of excavation behavior in poor
quality rock masses, Hoek and Brown (1997) modified the relation of Serafim and
Pereira (1983). Rowe and Armitage (1984) correlated the rock mass modulus deduced
from a large number of field tests of drilled shafts under axial loading with the average
unconfined compressive strength of weak rock deposits in which the drilled shafts are
founded. According to an extensive literature review, Heuze (1980) concluded that the
deformation modulus of rock masses ranges between 20 and 60% of the modulus
measured on intact rock specimens in the laboratory. Hoek and Brown (1980)
proposed an empirical failure criterion for rock masses using two parameters, m and s,
which are related to the degree of rock mass fracturing. Empirical expressions have
also been proposed between those parameters and RQD (Deere et al., 1967), RMR
(Bieniawski, 1974) and Q ratings (Barton et al., 1974).
2. The second indirect approach consists of treating jointed rock mass as an equivalent
anisotropic continuum with deformability and strength properties that are directional
and reflect the properties of intact rock and those of the discontinuity sets, i.e.,
orientation, spacing and normal and shear stiffnesses. The discontinuities are
characterized without reference to their specific locations. Singh (1973), Kulhawy
(1978), Gerrard (1982a, b, 1991), Amadei (1983), Oda et al. (1984), Fossum (1985),
Yoshinaka and Yambe (1986), Oda (1988), Chen (1989) and Amadei and Savage
(1993) have derived the deformation moduli (or compliances) for rock masses with
continuous persistent discontinuities by considering the load-deformation relation for
each component (intact rock and discontinuities) and assuming that the behavior of the
jointed rock mass is the summation of each component response. Kulatilake et al.
(1992, 1993) and Wang (1992) derived relationships between the deformation
parameters of rock masses with non-persistent discontinuities and fracture tensor

parameters based on DEM (discrete element method) analysis results of simulated
rock mass blocks. Jaeger (1960) and Jaeger and Cook (1979) presented an equilibrium
continuum strength model for jointed rock masses under axisymmetric loading
condition. In their model, the strength of both the intact rock and the discontinuities
are described by the Coulomb criterion. Since the effect of the intermediate principal
stress is not considered in the model of Jaeger (1960) and Jaeger and Cook (1979),
Amadei (1988) and Amadei and Savage (1989, 1993) derived solutions for the
strength of a jointed rock mass under a variety of multiaxial states of stress. As in the
model of Jaeger (1960) and Jaeger and Cook (1979), the modeled rock mass is cut by
a single discontinuity set. In the formulations of Amadei (1988) and Amadei and
Savage (1989, 1993), however, the intact rock strength is described by the Hoek-
Brown strength criterion and the discontinuity strength is modeled using a Coulomb
criterion with a zero tensile strength cut-off.
3. The third indirect approach is to treat discontinuities as discrete features. This is
usually done in numerical methods, such as the finite element (Goodman et al., 1968;
Ghaboussi et al., 1973; Desai et al., 1984), boundary element (Crouch & Starfield,
1983) and discrete element (Cundall, 1971; Lemos et al., 1985; Lorig et al., 1986;
Cundall, 1988; Hart et al., 1988) methods, in which the complex response of
discontinuities to normal and shear stresses can be introduced in an explicit manner.
The main drawbacks of this approach is that so far, due to computer limitations, only
rock masses with a limited number of discontinuities can be analyzed.
4.2 DEFORMABILITY AND STRENGTH OF ROCK

DISCONTINUITIES
The behavior of jointed rock masses is dominated by the behavior of discontinuities in

the rock mass. To consider the effects of discontinuities on the deformability and strength
of rock masses, the deformability and strength of rock discontinuities should be known
first.
4.2.1 Deformation behavior of rock discontinuities

The deformation properties of individual rock discontinuities are described by normal
stiffness kn and shear stiffness ks. These refer to the rate of change of normal stress σn and
shear stress τ with respect to normal displacement un and shear displacement us. Details
about the definition of kn and ks are presented in the following.
If a compressive normal stress σn is applied on a rock discontinuity, it would cause its
closure by a certain amount, say un. Figure 4.1(a) shows a typical relationship between σn
and un. The slope of the curve in Figure 4.1(a) gives the tangential normal stiffness kn of
the discontinuity and, at any stress level, is defined as
(4.1)
where ∆ denotes an increment.

It is noted that kn is small when σn is small but rapidly builds up as the discontinuity
closes. There is actually a limit of discontinuity closure and σn→∞ as this limit (unc) is
reached. Goodman et al. (1968) proposed a hyperbolic relationship given by
(4.2)
Fig. 4.1 Typical stress-relative

displacement relationship: (a) σn
versus un; and (b) τ versus us.
where α and β are constants defining the shape of the hyperbolic curve between σn and un.
Differentiating Equation (4.2), we obtain the expression for kn as
(4.3)
which can be rewritten as
(4.4)
It is noted that Equation (4.4) is valid for compressive normal stresses only. It is usual to
assume that discontinuities do not offer any resistance to tensile normal stresses implying
kn =0 if σn is tensile.
If a shear stress τ is applied on the discontinuity, there will be a relative shear
displacement us on the discontinuity. Figure 4.1(b) shows a typical relationship between τ
and us. It is now possible to define a tangential shear stiffness ks exactly in the same way
as was done for the normal stiffness. Thus
(4.5)
ks is roughly constant till a peak value of the shear stress is reached. Nonlinear values
can, however, be adopted if justified by experimental results.
It is noted that for discontinuities (especially rough discontinuities), an increment of a
shear stress can produce an increment of relative displacement in the normal direction
and vice versa an increment of a normal stress can produce an increment of relative
displacement in the shear direction. This behavior is called dilation of discontinuities. If
the relative shear displacement is broken into two components (along two perpendicular
coordinate axes s and t on the discontinuity plane—see Fig. 4.2), the general constitutive
relation for a discontinuity including the dilation behavior can be expressed as
(4.6)
where the subscripts ‘s’ and ‘t’ represent two orthogonal directions in the discontinuity
plane; the subscript ‘n’ represents the direction normal to the discontinuity plane; us and
ut are the shear displacements respectively in directions s and t; un is the closure
displacement; τs and τt are the shear stresses respectively in directions s and t; σn is the
normal stress; and [Cij] (i, j=s, t, n) is the compliance matrix of the discontinuity.
Elements of the compliance matrix can be found experimentally by holding two of the
stresses constant (for example at zero) and then monitoring the three relative
displacement components associated with changes in the third stress component (Priest,
1993).
For simplicity, the following assumptions are often made for the behavior of a single
discontinuity:
(1) Deformation behavior is the same in all directions in the discontinuity plane. Thus
Css=Ctt, Cst=Cts, Csn=Cta and Cns=Cnt.
(2) The dilation (coupling) effect is neglected, i.e., Cij (i≠j) in Equation (4.6) are zero.
With the above two assumptions, Equation (4.6) can be simplified to
(4.7)
where ks and kn are respectively the discontinuity shear and normal stiffness as described
above.
4.2.2 Shear strength of rock discontinuities

Discontinuities usually have negligible tensile strength and a shear strength that is, under
most circumstances, significantly smaller than that of the surrounding intact rock
material. The following describes several shear strength models for rock discontinuities.
Fig. 4.2 A local coordinate system s, t,

n.
(a) Coulomb model

The simplest shear strength model of discontinuities is the Coulomb failure criterion.
This model can be expressed by the well-known equation
(4.8)
where τ is the shear strength of the discontinuities; cj and are respectively the cohesion
and internal friction angle of the discontinuities; and σ′n is the effective normal stress on
the discontinuity plane. It need be noted that the” primes” for cj and have been omitted
for brevity although they are for the effective stress conditions.
(b) Bilinear shear strength model

Usually the shear stress-normal stress relation of the discontinuities is non-linear. Patton
(1966) addressed this problem by formulating the bilinear model as shown in Figure 4.3.
At normal stresses less than or equal to σ′0 the shear strength is given by
(4.9)
where is the basic friction angle for an apparently smooth surface of the rock material;
and i is the effective roughness angle. Barton and Choubey (1977) have listed values of
determined experimentally by a number of authors. Some representative values are
listed in Table 4.1.
At normal stresses greater than or equal to σ′0 the shear strength is given by
(4.10)
where ca is the apparent cohesion derived from the asperities; and is the residual
friction angle of the rock material forming the asperities.
Jaeger (1971) proposed the following shear strength model to provide a curved
transition between the straight lines of the Patton model
(4.11)
Table 4.1 Basic friction angles for different rock

materials (after Barton & Choubey, 1977).
Rock Types dry (degrees) wet (degrees)
Sandstone 26–35 25–34
Siltstone 31–33 27–31
Limestone 31–37 27–35

Basalt 35–38 31–36
Fine granite 31–35 29–31
Coarse granite 31–35 31–33
Gneiss 26–29 23–26
Slate 25–30 21
Fig. 4.3 Bilinear shear strength models

(Equations 4.9 and 4.10) with
empirical transitions curve (Equation
4.11).
where d is an experimentally determined empirical parameter which controls the shape of
the transition curve.
(c) Barton model

A direct, practical approach to predicting the shear strength of discontinuities on the basis
of relatively simple measurements was developed by Barton and his coworkers (Barton,
1976; Barton & Choubey, 1977; Barton & Bandis, 1990). According to the Barton model,
the shear strength τ of a discontinuity subjected to a normal stress σ′n in a rock material
with the basic friction angle is given by
(4.12)
where JRC is the discontinuity roughness coefficient; and JCS is the discontinuity wall
compressive strength.
The discontinuity roughness coefficient JRC provides an angular measure of the
geometrical roughness of the discontinuity surface in the approximate range 0 (smooth)
to 20 (very rough). The JRC can be estimated in a number of ways. Barton and Choubey
(1977) present a selection of scaled typical roughness profiles (Fig. 4.4), which facilitate
the estimation of JRC for real discontinuities by visual matching. Barton (1987)
published a table relating Jr (discontinuity roughness number in the Q classification
system) to JRC (see Fig. 4.5). Barton and Bandis (1990) suggest that JRC can also be
estimated from a simple tilt shear test in which a pair of matching discontinuity surfaces
are tilted until one slides over the other. The JRC can be back-figured from the tilt angle
α (Fig. 4.6) using the following equation:
(4.13)
Fig. 4.4 Typical discontinuity

roughness profiles and associated JRC
values (after Barton & Choubey,
1977).
Fig. 4.5 Relationship between Jr in Q-

system and JRC for 200 mm and 1 m
samples (after Barton, 1987).
The nail brush is one of the simple methods to record surface profiles. Tse and Cruden
(1979) present a method for estimating JRC based on a digitization of the discontinuity
surface into a total of M data points spaced at a constant small distance ∆x along the
profile. If yi is the amplitude of the ith data point measured above (yi+) and below (yi−)
the center line, the root mean square Z2 of the first derivative of the roughness profile is
given by
Fig. 4.6 Tilt test to measure the tilt

angle α (after Barton & Bandis, 1990).
(4.14)
By digitizing the ten typical roughness profiles presented in Figure 4.4 and then
conducting a series of regression analyses, Tse and Cruden (1979) found that there is a
strong correlation between JRC and Z2. On this basis, they proposed the following
expression for estimating JRC:
JRC≈32.2+32.47 log10Z2
(4.15)
The increasing availability of image analysis hardware and low-cost digitizing pads
makes the method of Tse and Cruden (1979) a valuable objective alternative for the
assessment of JRC. This approach should be used with caution, however, since Bandis et
al. (1981) have shown that both JRC and JCS reduce with increasing scale. The idea of
applying statistical and probabilistic analysis of surface profiles to the calculation of JRC
has recently been examined and extended by several authors, notably McWilliams et al.
(1990), Roberds et al. (1990), Yu and Vayssade (1990), and Zongqi and Xu (1990).
These last authors, noting that the value of JRC is dependent upon the sampling interval
along the profile, propose the following extension to Equation (4.15)
JRC≈AZ2−B
(4.16)
where the constants A and B depend on the sampling interval ∆x, taking values of 60.32
and 4.51, respectively, for an interval of 0.25 mm, 61.79 and 3.47 for an interval of 0.5
mm, and 64.22 and 2.31 for an interval of 1.0 mm. Lee et al. (1990), applying the concept
of fractals to discontinuity surface profiles, obtained an empirical relation linking the
fractal dimension D to the JRC value, as follows:
(4.17)
Unfortunately Lee et al. (1990) do not explain adequately how the fractal dimension D
should be determined in practice. Odling (1994) proposed a method for determining the
fractal dimension D. In Odling’s method, the roughness of a fracture surface is
represented by the structure function S. For a fracture surface profile, S is defined as
(4.18)
where M is the number of data points at a sampling interval ∆x, and yi is the amplitude of
the ith data point measured above (yi+) and below (yi−) the center line. The structure
function is thus simply the mean square height difference of points on the profile at
horizontal separations of ∆x. The structure function is related to the Hurst exponent H
(Voss, 1988; Poon et al., 1992):
S(∆x)=A(∆x)2H
(4.19)
Thus, if a log-log plot of S(∆x) versus ∆x gives an acceptably straight line, the slope of
this line gives 2H. A is an amplitude parameter and is equivalent to the mean square
height difference at a sampling interval of l unit, and is therefore dependent on the units
of measurement. From H, the fractal dimension can be determined from the following
equation (Voss, 1988):
D=E−H
(4.20)
where E is the Euclidean dimension of embedding medium. E=2 for surface profiles.
If the discontinuity is unweathered, JCS is equal to the unconfined compressive
strength of the rock material σc, determined by point load index tests or compression tests
on cylindrical specimens. If there has been softening or other forms of weathering along
the discontinuity, then JCS will be less than σc and must be estimated in some way.
Suggested methods for estimating JCS are published by ISRM (1978). Barton and
Choubey (1977) explain how the Schmidt hammer index test can be used to estimate JCS
from the following empirical expression
log10JCS≈0.88γR+1.01
(4.21)
where γ is the unit weight of the rock material (MN/m3), R is the rebound number for the
L-hammer and JCS has the units MPa in the range 20 to approximately 300 MPa.
Although the Schmidt hammer is notoriously unreliable, particularly for heterogeneous
materials, it is one of the few methods available for estimating the strength of a surface
coating of material.
Equation (4.12) suggests that there are three factors which control the shear strength of
rock discontinuities: the basic friction angle , a geometrical component JRC, and an
asperity failure component controlled by the ratio JCS/σ′n. Research results show that
both JRC (geometrical component) and JCS (asperity failure component) decrease with
increasing scale (Bandis, 1990; Barton & Bandis, 1982) (see Fig. 4.7). Based on
extensive testing of discontinuities, discontinuity replicas, and a review of literature,
Barton and Bandis (1982) proposed the scale corrections for JRC and JCS:
(4.22a)
(4.22b)
Fig. 4.7 Influence of scale on the three

components of the shear strength of a
rough discontinuity (after Bandis,
1990; Barton & Bandis, 1990).
where JRC0, JCS0 and L0 (length) refer to 100 mm laboratory scale samples and JRCn,
JCSn and Ln refer to in situ block sizes.
It is worth noting two important limitations on the use of Barton model for estimating
the shear strength of discontinuities. Barton and Choubey (1977) suggest that the curves
should be truncated such that the maximum allowable shear strength for design purposes
is given by arctan(τ/σ′n)=70°. For example, curve 1 in Figure 4.8 has a linear “cut-off’
representing the maximum suggested design value of 70° for the total frictional angle.
Barton (1976) cautioned that when the effective normal stress exceeds the unconfined
compressive strength of the rock material, the measured shear strength is always
appreciably higher than that predicted by Equation (4.12). Noting that this discrepancy
was probably due to the effect of confining stresses increasing the strength of asperities,
Barton
Fig. 4.8 Range of peak shear strength

for 136 joints representing eight
different rock types. Curves 1, 2 and 3
are evaluated using Equation (4.12)
(after Barton & Choubey, 1977).
proposed that a high stress version of Equation (4.12) could be obtained by replacing JCS
by (σ′1−σ′3), i.e.,
(4.23)
where σ′1 is the effective axial stress required to yield the rock material under an effective
confining stress σ′3. The failure stress σ′1 can either be determined experimentally or can
be estimated from an appropriate yield criterion such as the Hoek-Brown criterion.
(d) Comments
The following comments should be noted when using the shear strength criteria described
in the previous sections:
1. The Coulomb model is applicable to discontinuities with planar surfaces and the
bilinear model to discontinuities with rough surfaces. Since the discontinuity
roughness coefficient JRC is incorporated in the strength criterion, the Barton model is
applicable to discontinuities with either planar or rough surfaces.
2. The shear strength criteria described in the previous sections are applicable to
discontinuities in which rock wall contact occurs over the entire length of the surface
under consideration. The shear strength can be reduced drastically when part or all of
the surface is not in intimate contact, but covered by soft filling material such as clay
gouge.
4.3 DEFORMABILITY OF ROCK MASS
4.3.1 Definition of modulus

The modulus relates the change in applied stress to the change in the resulting strain.
Mathematically, it is expressed as the slope of a given stress-strain response. Since a rock
mass seldom behaves as an ideal linear elastic material, the modulus value is dependent
upon the proportion of the stress-strain response considered. Figure 4.9 shows a stress-
strain curve typical of an in-situ rock mass containing discontinuities with the various
moduli that can be obtained. Although the curve, as shown, is representative of a jointed
mass, the curve is also typical of intact rock except that the upper part of the curve tends
to be concaved downward at stress levels approaching failure. As can be seen in Figure
4.9 there are at least four portions of the stress-strain curve that can be used for
determining in-situ rock mass moduli: the initial tangent modulus, the elastic modulus,
the tangent recovery modulus, and the deformation modulus:
a. Initial tangent modulus. The initial tangent modulus is determined from the slope of a
line constructed tangent to the initial concave upward section of the stress-strain curve
(i.e. line 1 in Fig. 4.9). The initial curved section reflects the effects of discontinuity
closure in in-situ tests and micro-crack closure in tests on small laboratory specimens.
b. Elastic modulus. Upon closure of discontinuities/micro-cracks, the stress-strain curve

becomes essentially linear. The elastic modulus, frequently referred to as the modulus
of elasticity, is derived from the slope of this linear (or near linear) portion of the
curve (i.e. line 2 in Fig. 4.9). In some cases, the elastic modulus is derived from the
slope of a line constructed tangent to the stress-strain curve at some specified stress
level. The stress level is usually specified as 50 percent of the maximum or peak
stress.
c. Recovery modulus. The recovery modulus is obtained from the slope of a line
constructed tangent to the initial segment of the unloading stress-strain curve (i.e. line
3 in Fig. 4.9). As such, the recovery modulus is primarily derived from in-situ tests
where test specimens are seldom stressed to failure.
d. Deformation modulus. Each of the above moduli is confined to specific regions of the
stress-strain curve. The deformation modulus is determined from the slope of the
secant line established between zero and some specified stress level (i.e. line 4 in Fig.
4.9). The stress level is usually specified as half of the maximum or peak stress.
Since the actual jointed rock masses do not behave elastically, deformation modulus is
usually used in practice.
Fig. 4.9 Stress-strain curve typical of

in-situ rock mass with various moduli
that can be obtained.
4.3.2 Empirical methods for estimating rock mass deformation modulus

A number of empirical methods have been developed that correlate various rock quality
indices or classification systems to in-situ deformation modulus of rock masses. The
commonly used include correlations between the deformation modulus and RQD, RMR,
GSI and Q. The definition of RQD, RMR, GSI and Q and the methods for determining
them have been discussed in Chapter 2.
(a) Methods relating deformation modulus with RQD

Based on field studies at Dworshak Dam, Deere et al. (1967) suggested that RQD be used
for determining the rock mass deformation modulus. By adding further data from other
sites, Coon and Merritt (1970) developed a relation between RQD and the modulus ratio
Em/Er, where Em and Er are the deformation moduli respectively of the rock mass and the
intact rock (see Fig. 4.10).
Gardner (1987) proposed the following relation for estimating the rock mass
deformation modulus Em from the intact rock modulus Er by using a reduction factor αE
which accounts for frequency of discontinuities by RQD:
Fig. 4.10 Variation of Em/Er with RQD

(after Coon & Merritt, 1970).
Em=αEEr
(4.24a)
αE=0.0231(RQD)−1.32≥0.15
(4.24b)
This method is adopted by the American Association of State Highway and

Transportation Officials in the Standard Specification for HighwayBridges (AASHTO,
1989). For RQD> 57%, Equation (4.24) is the same as the relation of Coon and Merritt
(1970). For RQD< 57%, Equation (4.24) gives Em/Er=0.15.
It is noted that the RQD−Em/Er relations by Coon and Merritt (1970) and Gardner
(1987) have the following limitations (Zhang & Einstein, 2000b):
(1) The range of RQD<60% is not covered and only an arbitrary value of Em/Er can be
selected in this range.
(2) For RQD=100%, Em is assumed to be equal to Er. This is obviously unsafe in design
practice because RQD=100% does not mean that the rock is intact. There may be
discontinuities in rock masses with RQD=100% and thus Em may be smaller than Er
even when RQD=100%.
Zhang and Einstein (2000b) added further data collected from the published literature to
cover the entire range 0≤RQD≤100% (see Fig. 4.11). It can be seen that the data in
Figure 4.11 shows a large scatter, which may be caused by the following three factors
(Zhang & Einstein, 2000b):
(1) Testing Methods
The data in Figure 4.11 were obtained with different testing methods. For
example, Deere et al. (1967) used plate load tests while Ebisu et al. (1992) used
borehole jacking tests. Different testing methods may give different values of
deformation modulus even for the same rock mass. According to Bieniawski
(1978), even a single testing method, such as the flat jack test, can lead to a wide
scatter in the results even where the rock mass is very uniform.
(2) Discontinuity Conditions
RQD does not consider the discontinuity conditions, such as the aperture and
fillers. However, the discontinuity conditions have a great effect on the rock mass
deformation modulus. Figure 4.12 shows the variation of Em/Er with the average
discontinuity spacing s for different values of kn/Er using the Kulhawy
(1978)model(see Section 4.3.3) It can be seen that kn/Er which represents the
discontinuity conditions has a great effect on the rock mass deformation modulus.
Fig. 4.11 Em/Er−RQD data and

proposed Em/Er−RQD relations (after
Zhang & Einstein, 2000b).
(3) Insensitivity of RQD to Discontinuity Frequency

RQD used in Figure 4.11 is defined in terms of the percentage of intact pieces of
rock (or discontinuity spacings) greater than a threshold value t of 0.1 m.
According to Harrison (1999), the adoption of a threshold value t of 0.1 m leads
to the insensitivity of RQD to the change of discontinuity frequency λ or mean
discontinuity spacing s. As discussed in Chapter 3, for a negative exponential
distribution of discontinuity spacings, the theoretical RQD can be related to the
discontinuity frequency λ by Equation (3.18). Figure 4.13 shows the variation of
RQD with λ. It can be seen that, for a threshold value t of 0.1 m, when
discontinuity frequency λ increases from 1 m−1 to 8 m−1 (i.e., the mean
discontinuity spacing s decreases from 1 m to 0.125 m) RQD only decreases from
99.5% to 80.9%, which is a range of only 23%. However, when the mean
discontinuity spacing s decreases from 1 m to 0.125 m, the rock mass deformation
modulus will vary over a large range. As shown in Figure 4.12, with kn/Er=1,
Em/Er changes from 0.5 to 0.11 when s decreases from 1 m to 0.125 m. Harrison
(1999) showed that the sensitivity of RQD to the mean discontinuity spacing s is
closely related to the adopted threshold value t. For example, if a threshold value t
of 0.5 m is used, the corresponding RQD will change from 91.0% to 9.2% when
A, increases from 1 m−1 to 8 m−1 (see.Fig. 4.13).
Considering the data shown in Figure 4.11, Zhang and Einstein (2000b) proposed the
following relations between the rock mass deformation modulus and RQD:
Lower bound:
Em/Em=0.2×100.0186RQD−1.91
(4.25a)
Upper bound:
Em/Er=1.8×100.0186RQD−1.91
(4.25b)
Fig. 4.12 Variation of Em/Er with

average discontinuity spacing s for
different values of kn/Er using
Kulhawy (1978) model (after Zhang &
Einstein, 2000b).
Mean:
Em/Er=100.0186RQD−1.91
(4.25c)
The mean relation between Em/Er and RQD was obtained by regression of the data in
Figure 4.11. The coefficient of regression, r2, is 0.76. The upper bound could be put
somewhat higher but it was selected to be conservative.
RQD is a directionally dependent parameter and its value may change significantly,
depending on the borehole orientation. Therefore, it is important to know the borehole
orientation when estimating the rock mass deformation modulus Em using the
Em/Er−RQD relationship. To reduce the directional dependence of RQD, the relationship
suggested by Palmstrom (1982) [Equation (2.2) in Chapter 2] can be used to estimate
RQD.
(b) Methods relating deformation modulus with RMR or GSI

The empirical relationship between the RMR rating value and the in situ rock mass
modulus is shown in Figure 4.14. Bieniawski (1978) studied seven projects and suggested
the following equation to predict rock mass modulus from RMR:
Em=2RMR−100 (GPa)
(4.26)
Fig. 4.13 RQD—discontinuity

frequency relations for threshold
values of 0.1 and 0.5 m (after Harrison,
1999 but with different threshold
values).
The obvious deficiency of this equation is that it does not give modulus values for RMR
values less than 50. Additional studies carried out on rock masses with qualities ranging
from poor to very good indicated that the rock mass modulus could be related to RMR by
(Serafim & Pereira, 1983):
E=10(RMR−10)/40 (GPa)
(4.27)
Equation (4.27) has been found to work well for good quality rocks. However, for many
of the poor quality rocks it appears to predict deformation modulus values which are too
high (Hoek & Brown, 1997). Based on practical observations and back analysis of
excavation behavior in poor quality rock masses, Hoek and Brown (1997) modified
Equation (4.27) for unconfined compressive strength of intact rock σc<100 MPa as
follows:
(4.28)
where σc is in the unit of MPa. Note that GSI (Geological Strength Index) has been
substituted for RMR in Equation (4.28).
Fig. 4.14 Correlation between in situ

deformation modulus and RMR (after
Serafim & Pereira, 1983).
Johnston et al. (1980) also found that Equation (4.27) overestimates the rock mass
modulus for poor quality rocks. They reported that the results of various in situ load tests
in moderately weathered Melbourne mudstone of σc in the range 2 to 3 MPa yielded a
rock mass modulus of about 0.5 GPa for estimated RMRs of about 70. If we use Equation
(4.28) with σc=2.5 MPa and GSI=RMR−5=65, we can obtain an Em of 3.7 GPa which is
much closer to the measured value of about 0.5 GPa than the value of 31.6 GPa
calculated using Equation (4.27).
(c) Methods relating deformation modulus with Q

Barton et al. (1980) suggested the following relationships between in situ deformation
modulus values and Q values:
Lower bound:
Em=10loge Q (GPa)
(4.29)
Upper bound:
Em=40loge Q (GPa)
(4.29)
Mean:
Em=25loge Q (GPa)
(4.29)
where Q is the rock quality index as discussed in Chapter 2
(d) Methods relating deformation modulus with σc

Rowe and Armitage (1984) correlated the rock mass modulus deduced from a large
number of field tests of drilled shafts under axial loading with the average unconfined
compressive strength σc of weak rock deposits in which the drilled shafts was founded as
follows:
(4.30)
Radhakrishnan and Leung (1989) found good agreement between the rock mass moduli
obtained from back analysis of load-settlement relationship of large diameter drilled
shafts in weathered sedimentary rocks and those computed from Equation (4.30). It is
interesting to note that Equation (4.30) is equivalent to Equation (4.28) for GSI=23 which
corresponds to rock masses of very poor quality.
(e) Comments
Although the methods of this category are most widely used in practice, there are some
limitations:
1. The anisotropy of the rock mass caused by discontinuities is not considered.
2. Different empirical relations often give very different deformation moduli of rock
masses at the same site.
4.3.3 Equivalent continuum approach for estimating rock mass

deformation modulus
Equivalent continuum approach treats jointed rock mass as an equivalent anisotropic
continuum with deformability that reflects the deformation properties of the intact rock
and those of the discontinuity sets.
(a) Rock mass with persistent discontinuities

For rock masses with persistent discontinuities, analytical expressions for their
deformation properties have been derived by a number of authors, including Duncan and
Goodman (1971), Singh (1973), Kulhawy (1978), Gerrard (1982a, b, 1991), Amadei
(1983), Oda et al. (1984), Fossum (1985), Yoshinaka and Yambe (1986), Oda (1988) and
Amadei and Savage (1993). The basic idea used by different authors to derive the
expressions for deformation properties is essentially the same, i.e., the average stresses
are assumed to distribute throughout the rock mass and the overall average strains of the
rock mass are contributed by both the intact rock and the discontinuities. The only
difference is the method for determining the additional deformation due to the
discontinuities. Some of the typical results are presented in the following.
The three-dimensional equivalent continuum model presented by Kulhawy (1978) for
a rock mass containing three orthogonal discontinuity sets is shown in Figure 4.15. The
intact rock material is defined by the Young’s modulus Er and Poisson’s ratio νr, while
the discontinuities are described by normal stiffness kn, shear stiffness ks, and mean
discontinuity spacing s. For this model, the deformation properties of the equivalent
orthotropic elastic mass are given as
(4.31)
Fig. 4.15 Rock mass model of

Kulhawy (1978).
(4.32)
(4.33)
for i=x, y, z with j=y, z, x and k=z, x, y. These equations describe the rock mass elastic
properties completely. The single discontinuity model is a special case of the foregoing in
which sx=sy=∞. Singh (1973), Amadei (1983), Chen (1989) and Amadei and Savage
(1993) obtained the same expressions as above for deformation properties of rock masses
containing three orthogonal discontinuity sets.
For engineering convenience, it is useful to define a modulus reduction factor, αE,
which represents the ratio of the rock mass to rock material modulus. This factor can be
obtained by re-writing Equation (4.31) as
(4.34)
The relationship is plotted in Figure 4.16. This figure shows smaller values of αE in rock
masses with softer discontinuities (larger Er/kn values).
Unfortunately, the mean discontinuity spacing is not easy to obtain directly and, in
normal practice, RQD values are determined instead. Using a physical model, the RQD
can be correlated with the number of discontinuities per 1.5 meters (5 ft) core run, a
common measure in practice. This relationship is shown in Figure 4.17. Combining
Figures 4.16 and 4.17 yields Figure 4.18, which relate αE and RQD with Er/kn as an
additional parameter.
Fig. 4.16 Modulus reduction factor

versus discontinuity spacing (after
Kulhawy, 1978).
Fig. 4.17 RQD versus number of

discontinuities per 1.5m run (after
Kulhawy, 1978).
Fig. 4.18 Modulus reduction factor

versus RQD (after Kulhawy, 1978).
Consider the jointed rock under uniaxial loading (as shown in Fig. 4.19). The constitutive
relation in the n, s, t coordinate system can be defined from the single discontinuity
model of Kulhawy (1978). In the global coordinate system x, y, z, the constitutive relation
can be determined using second tensor coordinate transformation rules. In matrix form
this gives (Amadei & Savage, 1993).
(ε)xyz=(A)xyz(σ)xyz
(4.35)
where and .
The components aij=aji (i, j=1−6) of the compliance (A)xyz depend on the dip angle θ as
follows:
(4.36a)
(4.36b)
(4.36c)
Fig. 4.19 Jointed rock under uniaxial

loading (after Amadei & Savage,
1993).
(4.36d)
(4.36e)
(4.36f)
(4.36g)
(4.36h)
(4.36i)
(4.36j)
(4.36k)
All other coefficients aij vanish. Note that for the orientation of the discontinuities
considered here, the jointed rock has a plane of elastic symmetry normal to the z-axis. If
the discontinuity set is inclined with respect to x and z axes or if the rock sample under
consideration has two or three orthogonal discontinuity sets, then new expressions must
be derived.
Gerrard (1982a, b, 1991) presented an approximate method for determining the
equivalent elastic properties for a rock mass containing several sets of discontinuities. His
analysis is based on the assumption that the strain energy stored in the equivalent
continuum is the same as that stored in the discontinuous system. The first step is to rank
the various discontinuity sets according to their mechanical significance. Taking the least
significant set first, a compliance matrix for the equivalent continuum is determined. This
equivalent continuum is then regarded as the anisotropic ‘rock material’ for the next
discontinuity set, and so on until all discontinuity sets have been incorporated. A rotation
matrix must be applied to transform the equivalent continuum compliance matrix from
local coordinate axes, associated with one discontinuity orientation, to axes associated
with the next. The models for one, two and three sets of discontinuities are briefly
described in the following:
1) A single set of discontinuities can be modeled by considering a system of alternating
layers of approximately equal spacing. The interfacing planes are perpendicular to the
z axis of the coordinate set x, y, z. Material ‘a’ represents the rock material with
thickness Ta, material ‘b’ the discontinuity material with thickness Tb, and material ‘c’
is the homogeneous material equivalent to the system of alternating layers of ‘a’ and
‘b’ (see Fig. 4.20). The properties of material ‘c’ can be determined by using a series
of equations which are not listed here because they are too cumbersome (Gerrard,
1982a).
2) A second set of planar parallel discontinuities can be incorporated, in this case the
discontinuities being perpendicular to the x-axis. Alternating layers of the equivalent
material ‘c’, with thickness Tc, and the discontinuity material ‘d’, with thickness Td,
taken together can be represented by the equivalent homogeneous material (see Fig.
4.21). In order that material ‘c’ behaves in an effectively homogeneous fashion when
it is incorporated into material ‘e’ it is necessary that Tc»Ta+Tb.
3) The third set of planar parallel discontinuities are perpendicular to the y-axis. In this
case the homogeneous equivalent material ‘g’ can represent the alternating layers of
equivalent material ‘e’ with thickness Tc and the discontinuity material ‘f with
thickness Tf (see Fig. 4.22). In this case, to ensure that material ‘e’ behaves in an
effectively homogeneous fashion when it is incorporated into material ‘g’ it is
necessary that Te»Tc+Td.
Fossum (1985) derived a constitutive model for a rock mass that contains randomly
oriented discontinuities of constant normal stiffness kn and shear stiffness ks. He assumed
that if the discontinuities are randomly oriented, the mean discontinuity spacing would be
the same in all directions taken through a representative sample of the mass. Arguing that
the mechanical properties of the discontinuous mass would be isotropic, Fossum derived
the following expressions for the bulk modulus Km and shear modulus Gm of the
equivalent elastic continuum:
(4.37)
Fig. 4.20 One dimensional system of

discontinuities (after Gerrard, 1982a).
Fig. 4.21 Two dimensional system of

(4.38)
The equivalent Young’s modulus and Poisson’s ratio can be obtained from
(4.39)
(4.40)
Fig. 4.22 Three dimensional system of

discontinuities: (a) Representation in
the x-y plane; (b) Oblique view of
At large values of mean discontinuity spacing s the equivalent modulus Em and Poisson’s
ratio νm approach the values Er and νr for the intact rock material. At very small values of
mean discontinuity spacing the equivalent modulus Em and Poisson’s ratio νm are given
by the following expressions
(4.41)
(4.42)
Considering the fact that the available methods do not consider the statistical nature of
jointed rock masses, Dershowitz et al. (1979) present a statistically based analytical
model to examine rock mass deformability. The statistical model is shown in Figure 4.23.
The rock is taken as a three dimensional circular cylinder. Deformation is assumed to
accrue both from the elasticity of intact rock and from displacement along discontinuities.
Displacements along intersecting discontinuities are assumed to be independent. In this
model compatibility of lateral displacements across jointed blocks is approximated by
constraining springs. Inputs to the model include stififness and deformation moduli,
stress state, and discontinuity geometry. Intact rock deformability is expressed by
Young’s modulus Er, set at 200,000 kg/cm2, a typical value. Discontinuity stiffnesses are
represented by normal stiffness kn set at 1,000,000 kg/cm3, and shear stiffness ks set at
200,000 kg/cm3. The stress state is described by vertical major principal stress σ1,
horizontal “confining” stress σ3. “Confining” stress σ3 is determined from initial stress σ30
and a spring constant kg as follows
σ3=σ30+kgδy
(4.43)
where δy is the calculated horizontal displacement; σ30 is set to 50 kg/cm2; and kg is set at
2500 kg/cm3, a value chosen to maximize the increase of stress with lateral strain without
causing rotation of principal planes.
Discontinuity geometry is described by three parameters: the mean spacing sm, the
mean orientation θm and the dispersion according to the Fisher model κ. Spacing is
assumed to follow an exponential distribution and orientation a Fisher distribution (Table
4.2).
Table 4.2 Distribution assumptions for deformation
model (after Dershowitz et al., 1979).
Discontinuity Distribution form
property
Spacing Exponential: λe−λs, λ=(mean spacing)−1
Size (Persistence) Completely persistent
Orientation
Normal stiffness Deterministic

Shear stiffness Deterministic
Some of the results are shown in Figures 4.24 to 4.27. The results show that the proposed
model is consistent with the results of Deere et al. (1967) and Coon and Merritt (1970)
(see Fig. 4.10), to the extent that the relationships between deformation and RQD are of
similar form.
The model proposed by Dershowitz et al. (1979) has the following limitations:
1) The analysis applies only to “hard” rock. Shears and weathering can only be
accommodated through changes in discontinuity stiffnesses, which is inadequate.
2) The analysis is for infinitesimal strains. Finite strains would violate the assumption of
independence among discontinuity displacements.
3) The analysis is for a homogeneous deterministic stress field specified extraneous to the
discontinuity pattern. Real rock masses may have complex stress distributions strongly
influenced by the actual jointing pattern.
4) Boundary conditions are highly idealized.
Fig. 4.23 Statistical model for jointed

rock (after Dershowitz et al., 1979).
Fig. 4.24 Relationship between Em/Er

and RQD, parallel discontinuities
(after Dershowitz et al., 1979).
(b) Rock mass with non-persistent discontinuities

For rock masses with non-persistent discontinuities, relationships between the
deformation properties and the fracture tensor parameters in two and three dimensions
have been derived by Kulatilake et al. (1992, 1993) and Wang (1992) from the discrete
element method (DEM) analysis results of generated rock mass blocks. The procedure
used to evaluate the effect of discontinuities and the obtained relationships between the
deformation properties and the fracture tensor parameters in three dimensions are
outlined in the following.
Fig. 4.25 Relationship between the

mean of Em/Er, E[Em/Er], and the mean
of RQD, E[RQD], subparallel
discontinuities distributed according to
Fisher (after Dershowitz et al., 1979).
Fig. 4.26 Relationship between the

standard deviation of Em/Er,
SD[Em/Er], and the mean of RQD,
E[RQD], subparallel discontinuities
distributed according to Fisher (after
Dershowitz et al., 1979).
The procedure for evaluating the effect of discontinuities on the deformability of rock
masses is shown in Figure 4.28. The first step is the generation of non-persistent
discontinuities in 2 m cubical rock blocks. The discontinuities were generated in a
systematic fashion as follows:
Fig. 4.27 Effect of stiffness values on

modulus ratio Em/En parallel
discontinuities (after Dershowitz et al.,
1979).
1) In each rock block, a certain number of discontinuities having a selected orientation

and a selected discontinuity size were placed to represent a discontinuity set.
2) Discontinuities were considered as 2D circular discs.
3) Discontinuity center locations were generated according to a uniform distribution.
4) Either a single discontinuity set or two discontinuity sets were included in each rock
block.
The generated discontinuity networks in the rock blocks are given in Table 4.3.
Table 4.3. Generated discontinuity networks of
actual discontinuities in the rock block for 3D DEM
analysis (after Kulatilake et al., 1992, 1993).
Number of Orientation Discontinuity Number of Discontinuity
discontinuity sets α/β size/ block size discontinuities location
One set 60°/45° 0.1–0.9 with step 5, 10, 20, 30 Uniform
0.1
94.42°/37.89° 0.3, 0.5, 0.6, 0.7, 5, 10, 20, 30 distribution
0.9
30°/45° 0.3, 0.5, 0.6, 0.7, 5, 10, 20,
0.8, 0.9
90°/45° 0.3, 0.5, 0.6, 0.7, 5, 10, 20
0.8, 0.9
68.2°/72.2° 0.3, 0.6, 0.7, 0.8 5, 10, 20, 30
248.9°/79.8° 0.3, 0.6, 0.7,0.8 5, 10, 20, 30
Two sets 60°/45° 0.1, 0.2, 0.3, 0.5, 10
0.6, 0.7
240°/60° 10
Fig. 4.28 Procedure for evaluating the

effect of discontinuity geometry
parameters on the deformability
properties of jointed rock (after
Kulatilake et al., 1993).
The second step is the generation of fictitious discontinuities according to the actual
nonpersistent discontinuity network generated in the rock block. In order to use the DEM
for 3D analyses of a generated rock block, the block should be discretized into polyhedra.
Since a typical non-persistent discontinuity network in 3D may not discretize the block
into polyhedra, it is necessary to create some type of fictitious discontinuities so that
when they are combined with actual discontinuities, the block was discretized into
polyhedra. Before the generation of fictitious discontinuities, the actual disc-shaped
discontinuities are converted into square-shaped ones having the same area. In order for
the fictitious discontinuities to simulate the intact rock behavior, an appropriate
constitutive model and associated parameter values for the fictitious discontinuities have
to be found. From the investigation performed on 2D rock blocks, Kulatilake et al. (1992)
found that by choosing the mechanical properties of the fictitious discontinuities in the
way given below, it is possible to make the fictitious discontinuities behave as the intact
rock:
Table 4.4. Values for the mechanical parameters of
intact rock, actual and fictitious discontinuities used
by Kulatilake et al. (1992, 1993) and Wang (1992).
Intact rock or Discontinuities Parameter Assigned value
Intact rock Young’s modulus Er 60 GPa
Poisson’s ratio νr 0.25
Cohesion cr 50 MPa
Tensile strength tr 10 MPa
Friction coefficient 0.839
Fictitious discontinuities Normal stiffness kn 5000 GPa/m

Shear stiffness ks 2000 GPa/m
Cohesion cj 50MPa
Dilation coefficient dj 0
Tensile strength tj 10 MPa
0.839
Friction coefficient
Actual discontinuities Normal stiffness kn 67.2 GPa/m
Shear stiffness ks 2.7 GPa/m
Cohesion cj 0.4 MPa
Tensile strength tj 0
Friction coefficient 0.654
(a) The strength parameters of the fictitious discontinuities are the same as those of the
intact rock.
(b) Gr/ks=0.008–0.012.
(c) kn/ks=2–3, with the most appropriate value being Er/Gr.
For the intact rock (granitic gneiss) studied by Kulatilake et al. (1992, 1993) and Wang
(1992), the approximate parameters of the fictitious discontinuities are shown in Table
4.4. The mechanical parameters of the actual discontinuities used by them are also shown
in this table. The constitutive models used for the intact rock and discontinuities (both
actual and fictitious) are shown in Figures 4.29 and 4.30, respectively.
The third step is the DEM analysis of the rock block (using the 3D distinct element
code 3DEC) under different stress paths and the evaluation of the effect of discontinuities
on the deformation parameters of the rock mass. In order to estimate different property
values of the jointed rock block, Kulatilake et al. (1993) and Wang (1992) used the
following stress paths:
1) The rock block was first subjected to an isotropic compressive stress of 5 MPa in three
perpendicular directions (x, y, z); then, for each of the three directions, e.g. the z-
direction, the compressive stress σz was increased, while keeping the confining
stresses in the other two directions (σx and σy) the same, until the failure of the rock
occurred (see Fig. 4.31). From these analysis results, it is possible to estimate the
deformation modulus of the rock block in each of the three directions and the related
Poisson’s ratios.
Fig. 4.29 Constitutive model assumed

for intact rock: (a) stress versus strain;
(b) Coulomb failure criterion with a
tension cut-off (after Kulatilake et al.,
1993).
2) The rock block was first subjected to an isotropic compressive stress of 5 MPa in three
perpendicular directions (x, y, z); then, on each of the three perpendicular planes, e.g.
the x-y plane, the rock was subjected to an increasing shear stress as shown in Figure
4.32. These analysis results can be used to estimate the shear modulus of the rock
block on each of the three perpendicular planes.
In the DEM analysis, during the loading process, displacements were recorded
simultaneously on each block face in the direction(s) needed to calculate the required
block strains. On each block face, five points were selected to record the displacement.
The average value of these five displacements was considered as the mean displacement
of this face for block strain calculations.
To make it possible to estimate the deformation properties of the rock block from the
DEM analysis results, Kulatilake et al. (1993) and Wang (1992) assumed that the rock
block was orthotropic in the x, y, z directions, regardless of the actual orientations of the
discontinuities, i.e.,
With the above constitutive model, the deformation moduli Ex, Ey, Ez and Poisson’s ratios
νxy, νxz, vyx, νyz, νzx, vzy can be estimated from the DEM analysis results of rock blocks
under stress path 1 (Fig. 4.31). The shear moduli Gxy, Gxz and Gyz can be estimated from
the DEM analysis results of rock blocks under stress path 2 (Fig. 4.32).
To reflect the effect of discontinuity geometry parameters on the deformation
properties, Kulatilake, et al. (1993) and Wang (1992) used the fracture tensor defined by
Oda (1982) as an overall measure of the discontinuity parameters—discontinuity density,
orientation, size and the number of discontinuity sets. For thin circular discontinuities, the
general form of the fracture tensor at the 3D level for the kth discontinuity set can be
expressed as (see also Chapter 3 about the discussion of fracture tensors)
(4.45)
where ρ is the average number of discontinuities per unit volume (discontinuity density),
r is the radius of the circular discontinuity (discontinuity size), n is the unit vector normal
to the discontinuity plane, f(n, r) is the discontinuity probability density function of n and
r, Ω/2 is a solid angle corresponding to the surface of a unit hemisphere, and ni and nj (i,
j=x, y, z) are the components of vector n in the rectangular coordinate system considered
(see Fig. 4.33). The solid angle dΩ is also shown in Figure 4.33. If the distributions of the
size and the orientation of the discontinuities are independent of each other, Equation
Fig. 4.30 Constitutive model assumed

for joints: (a) normal stress versus
normal displacement; (b) shear stress
versus shear displacement; and (c)
Coulomb failure criterion with a
tension cut-off (after Kulatilakeetal.,
1993).
Fig. 4.31 Stress paths of first type used

to perform DEM analysis of generated
rock blocks (after Kulatilakeetal.,
1993).
Fig. 4.32 Stress paths of second type

used to perform DEM analysis of
generated rock blocks (after
Kulatilakeetal, 1993).
(4.44)
(4.45) can be rewritten as follows
(4.46)
where f(n) and f(r) are the probability density functions of the unit normal vector n and
size r, respectively. If there are more than one discontinuity set in the rock mass, the
fracture tensor for the rock mass can be obtained by
(4.47)
wher N is the number of discontinuity sets in the rock mass. Fracture tensor Fij can also
be written in matrix form as follows
(4.48)
Fig. 4.33 Unit sphere used to define

the solid angle dΩ (after Oda, 1982).
Since the diagonal components of the fracture tensor Fxx, Fyy and Fzz express the
combined effect of discontinuity density and discontinuity size in the x, y and z
directions, respectively, Kulatilake et al. (1993) and Wang (1992) showed the obtained
deformation properties as in Figures 4.34 and 4.35. Putting the data in Figures 4.34(a)–(c)
and Figures 4.35(a)–(c) respectively together, Figures 4.36 and 4.37 were obtained,
which show that the deformation properties of jointed rock masses are related to the
corresponding components of the fracture tensor. As for the Poisson’s ratios of the
generated rock blocks, Kulatilake et al. (1993) and Wang (1992) found that they were
between 50 and 190% of the intact rock Poisson’s ratio.
(c) Comments
In the equivalent continuum approach, the elastic properties of the equivalent material are
essentially derived by examining the behavior of two rock blocks having the same
volume and by using an averaging process. One volume is a representative sample of the
rock
Fig. 4.34 Relations between rock block

deformation moduli and fracture tensor
components for different discontinuity
networks: (a) Ez/Er vs Fzz; (b) Ey/Er vs
Fyy; and (c) Ex/Er vs Fxx (after

shear moduli and summation of
corresponding fracture tensor
components for different discontinuity
networks: (a) Gxy/Gr vs (Fxx+Fyy); (b)
Gxz/Gr vs (Fxx+Fzz); and (c) Gyz/Gr vs
(Fyy+Fzz) (after Kulatilake et al.,
1993).

deformation modulus in any direction
Em and the fracture tensor components
in the same direction (after Kulatilake
et al., 1993).

shear modulus on any plane Gm and
the summation of fracture tensor
components on that plane (after
mass whereas the second volume is cut from the equivalent continuum and is subject to
homogeneous (average) stresses and strains. Therefore, the equivalent continuum
approach requires that the representative sample of the rock mass be large enough to
contain a large number of discontinuities. On the other hand, the corresponding
equivalent continuum volume must also be sufficiently small to make negligible stress
and strain variations across it. This leads to a dilemma which is typical in modeling
continuous or discontinuous composite media.
Numerous authors have used the equivalent continuum approach and derived the
expressions for the equivalent continuum deformation properties. Most of these
expressions are based on the assumption that the discontinuities are persistent. This is a
conservative assumption since, in reality, most of the discontinuities are non-persistent
with finite size.
For a rock mass containing non-persistent discontinuities, Kulatilake et al. (1992,
1993) and Wang (1992) derived relationships between the deformation properties and the
fracture tensor parameters from the DEM analysis results of generated rock mass blocks.
However, there exist limitations for the method they used and thus for the relationships
they derived as follows:
1. The generated rock mass block is assumed to be orthotropic in the x, y, z directions,
regardless of the actual orientations of the discontinuities. The appropriateness of this
assumption is questionable. For example, the two blocks shown in Figure 4.38 have
the same fracture tensor Fij, block 1 containing three orthogonal discontinuity sets
while block 2 containing 1 discontinuity set. It is appropriate to assume that block 1 is
orthotropic in the x, y, z directions. However, it is obviously inappropriate to assume
that block 2 is orthotropic in the x, y, z directions.
2. To do DEM analysis on the generated rock mass block, fictitious discontinuities are
introduced so that when they are combined with actual discontinuities, the block is
discretized into polyhedra. To make the fictitious discontinuities behave as the intact
rock, appropriate mechanical properties have to be assigned to the fictitious
discontinuities. From the investigation performed on 2D rock blocks, Kulatilake et al.
(1992) found a relationship between the mechanical properties of the fictitious
discontinuities and those of the intact rock. However, even if the mechanical
properties of the fictitious discontinuities are chosen from this relationship, the
fictitious discontinuities can only approximately behave as the intact rock. So the
introduction of fictitious discontinuities brings further errors to the final analysis
results.
3. Discontinuity persistence ratio PR (defined as the ratio of the actual area of a
discontinuity to the cross-section area of the discontinuity plane with the rock block)
should have a great effect on the deformability of rock masses. However, the
relationships derived by Kulatilake et al. (1992, 1993) and Wang (1992) does not
show any effect of PR on the deformability of jointed rock masses.
4. The conclusion that Ei/Er (i=x,y,z) is related only to Fii (i=x,y,z) is questionable. This
can be clearly seen from the two rock blocks shown in Figure 4.39. The two blocks
have the same fracture tensor component Fzz. From Figure 4.36, the two blocks will
have the same deformation modulus in the z-direction. However, block 2 is obviously
more deformable than block 1 in the z-direction.
Fig. 4.38 Two rock blocks having the

same fracture tensor but different joint
sets: (a) Rock block with three
orthogonal joint sets; and (b) Rock
block with one joint set.
Fig. 4.39 Two rock blocks having the

same fracture tensor component in z-
direction but different joint
orientations: (a) Rock block with joint
normal parallel to z-axis; and (b) Rock
block with joint normal inclined from
z-axis.
4.3.4 Direct consideration of discontinuities in numerical analysis

The direct consideration of discontinuities as discrete features is usually done in
numerical methods, such as the finite element, boundary element and discrete element
methods. Considering the fact that the finite element method (FEM) is the most widely
used numerical method in foundation analysis and design, only the methods for
representing discontinuities in finite element modeling are described in the following.
The presence of rock discontinuities is considered in finite element analyses by
employing special joint elements which describe the localized response of the
discontinuities. The various joint elements can be grouped into three general classes
(Curran & Ofoegbu, 1993):
(1) joint elements which use the nodal displacements as the independent degrees of
freedom (Goodman et al., 1968);
(2) joint elements which use the relative nodal displacements as the independent degrees
of freedom (Ghaboussi et al., 1973);
(3) thin-layer continuum elements assigned the behavior of discontinuities (Zienkiewicz
et al., 1970; Desai et al., 1984).
These classes of joint elements are briefly discussed below.
(a) Joint elements using nodal displacements as independent degrees-of-

freedom
This approach to modeling discontinuities was originally proposed by Goodman et al.
(1968) and is still commonly used today. A good summary of the development of the
element equations is given in Pande et al. (1990). The basic geometry of the element, for
2D problems, is illustrated in Figure 4.40(a). It has a length L (along the s-axis, i.e. the
discontinuity plane) and zero thickness (in the n-axis direction, i.e. normal to the
discontinuity plane). It is a four-node element, nodes 1 and 2 lying on the bottom surface
(subscript B), while nodes 3 and 4 lie on the top surface (subscript T). The relative
displacements ws (shear) and wn (normal) are given by
(4.49)
where the absolute displacements in s and n directions are denoted as u and ν,

respectively. Assuming that displacements vary linearly along each boundary and with
nodes 1 and 4 at s=−L/2 and nodes 2 and 3 at s=L/2 [Fig. 4.40(a)], the displacements at
the bottom and top boundaries are respectively given by
(4.50)
(4.51)
Fig. 4.40 Joint elements based on: (a)

nodal displacements; (b) relative nodal
displacements; and (c) thin-layer solid.
Substituting Equations (4.50) and (4.51) into Equation (4.49), the following can be
obtained
(4.52)
where α=1−2s/L, β=1+2s/L and the vector on the right-hand side is the node displacement
vector. For the Goodman model, the vector of the nodal force F is related to the relative
displacements w through the equation
(4.53)
where Ks and Kn are the shear and normal stiffness, respectively. Using the minimum
energy principle, the equilibrium equation for the element can be obtained in the form
Ku=F
(4.54)
where u is the vector of the nodal displacements in Equation (4.52), and
(4.55)
The element’s contribution Kg to the global stiffness matrix is given by
(4.56)
where θ is the angle measured anti-clockwise from the discontinuity local s-axis to the
global x-axis.
The following remarks can be made about Goodman’s joint element:
1) In the derivation above, the properties of discontinuities are assumed to be represented
by stiffness of discontinuities Ks and Kn, The stiffness matrix of the discontinuity
has no off-diagonal terms. This implies that there is no dilatancy of

discontinuities and the normal and shear behavior are uncoupled.
2) It is possible to formulate higher-order joint elements on the basis of Goodman’s joint
element. A procedure of numerical integration will have to be adopted as direct
integration is quite cumbersome.
3) Mehtab and Goodman (1970) have extended the formulation of the joint element
suitable for three-dimensional analysis. The joint element is a two-dimensional eight-
node quadrilateral with the nodes in the thickness direction being coincident.
(b) Joint elements using relative nodal displacements as independent

degrees-of-freedom
In this model, introduced by Ghaboussi et al. (1973), the joint element has a finite
thickness t and its degrees of freedom are the relative displacements ws and wn, which
vary linearly from s=−L/2 to s=L/2 [L is the length of the joint element and the local
coordinate system in Fig. 4.40(b) is in effect]. The joint strains εs (shear) and εn (normal)
are given by
(4.57)
Hence, the strain can be related to the relative displacements of the element as follows
(4.58)
where subscripts 1 and 2 stand for the two ends of the element, end 1 being at s=−L/2 and
end 2 at s=L/2. The stress-strain relation for the discontinuity is given by
(4.59)
where σs and σn are the shear and normal stresses respectively, and the 2×2 matrix D
represents the discontinuity stiffness.
The stiffness matrix K for the joint element in the local coordinate system is given by
K=∫BTDBdv
(4.60)
where B is the 2×4 matrix in Equation (4.58). The element’s contribution Kg to the global
stiffness matrix can be obtained using Equation (4.56).
Wilson (1977) further developed the technique of using relative displacements for the
joint element, including the expansion from two dimensions to three dimensions.
(c) Thin-layer elements

The two classes of joint elements described above differ from solid elements in some
fundamental ways, such as structural stiffness matrix, nature of the stress and strain
vectors and the strain-displacement relations. Because of these differences, their
incorporation into a regular finite element program (which is usually designed for solid
elements) requires significant modifications in the code. Desai et al. (1984) proposed the
thin-layer element as a means of reducing this problem.
The thin-layer element is basically a solid element, but its properties are assigned in
such way that its behavior closely approximate that of a discontinuity. A typical thin-
layer element is shown in Figure 4.40(c). This example is a six-node element but a four-
node element is acceptable. The stress-strain relations are derived in exactly the same
way as for other solid elements.
The main issue for using thin-layer elements in FEM analysis is choosing appropriate
material properties and thickness of the element. Desai et al. (1984) originally proposed
the thin-layer element mainly for applications in soil-structure interaction problems.
Since the interface is surrounded by the geological (soil) and structural materials, Desai et
al. (1984) proposed that the normal stiffness (i.e., deformation modulus in the direction
normal to the element plane) of the thin-layer element be chosen according to the
properties of the interface zone and the structural and geological materials, i.e.,
En=λ1(En)i+λ2(En)g+λ3(En)st
(4.61)
where (En)i, (En)g and (En)st are respectively the deformation modulus of the interface
zone and the geological and structural materials; and λ1, λ2 and λ3 are the participation
factors varying from 0 to 1. In a series of soil-structure interaction examples, Desai and
his coworkers chose λ2=λ3=0 and λ1=1 and obtained satisfactory results by assigning the
interface zone the same properties as the geological material. Desai et al. (1984) proposed
using shear testing devices [Fig. 4.41(a)] to obtain the shear modulus of the thin-layer
element. The expression used for obtaining a tangent shear modulus is given by
(4.62)
where t is the thickness of the element [Fig. 4.41(b)] and us is the relative displacement.
Fig. 4.41 Behavior at interface: (a)

schematic of direct shear test; and (b)
deformation at interface.
Fishman et al. (1991) used thin-layer elements for modeling rock discontinuities. Arguing
that the discontinuity interface is smooth and thus the normal deformation will be small,
they chose the Young’s modulus for the thin layer elements to be 10 times higher than
that of the surrounding solid rock elements. As for the shear modulus, they used the
method suggested by Desai et al. (1984) as described above.
The thickness of the thin-layer element has a great effect on the quality of simulation
of the interface (discontinuity) behavior. If the thickness is too large in comparison with
the dimension of the surrounding elements, the thin-layer element will behave essentially
as a solid element. If it is too small, computational difficulties such as numerical ill-
conditioning may arise. Investigations by Desai et al. (1984) suggest that values of t/B (B
is the smaller of the other two dimensions of the element) in the range 0.01≤t/B≤0.1 are
likely to give good results. In applying thin-layer elements for numerical modeling
jointed rock masses, Fishman et al. (1991) used thin-layer elements with
t/B=0.018−0.054 and got satisfactory results.
(d) Comments
Special joint elements have been used widely in the area of soil/rock and structure
interaction. The first two classes of joint elements as discussed in (a) and (b) differ from
solid elements in some fundamental ways, such as structural stiffness matrix, nature of
the stress and strain vectors and the strain-displacement relations. However, thin-layer
element is basically a solid element with a small thickness and a particular constitutive
relationship. Investigations by Ng et al. (1997) revealed that all these joint elements have
limitations, such as the problems of numerical ill-conditioning: if the joint elements have
a large aspect ratio (ratio of length to thickness), small values of the coefficients in the
diagonal of the stiffness matrices can create problems in the solution routine with a loss
in accuracy.
It happens very often that there is filling in rock discontinuities. Since the filling itself
is physically a solid, it is obviously more appropriate to use thin-layer elements than to
use the first two classes of joint elements to represent them in the FEM analysis. To use
thin-layer elements to represent discontinuities in the FEM analysis, appropriate
mechanical properties should be assigned to them. For 2D thin-layer elements, Desai et
al. (1984) proposed a procedure for determining the shear modulus and gave a general
idea (no detailed procedure) of evaluating the normal deformation modulus. Since, to
date, thin-layer elements have been used basically in 2D problems, no detailed
suggestions about selecting the properties of 3D thin-layer element are available.
4.4 STRENGTH OF ROCK MASS
4.4.1 Empirical strength criteria for rock mass

Several empirical strength criteria for rock masses have been formulated based on large-
scale testing and/or application experience and analysis. In the following, four typical
empirical strength criteria are described and discussed. Since the Hoek-Brown criterion is
the mostly widely used one, it is described and discussed in more details than the others.
(a) Hoek-Brown criterion

The Hoek-Brown criterion was originally published in 1980 (Hoek & Brown, 1980) and
has evolved to being used under conditions which were not visualized when it was
originally developed.
For intact rock, the Hoek-Brown criterion may be expressed in the following form
(4.63)
where σc is the uniaxial compressive strength of the intact rock material; σ′1 and σ′3 are
respectively the major and minor effective principal stresses; and mi is a material constant
for the intact rock. mi depends only upon the rock type (texture and mineralogy) as
tabulated in Table 4.5.
For jointed rock masses, the most general form of the Hoek-Brown criterion, which
incorporates both the original and the modified form, is given by
(4.64)
Table 4.5 Values of parameter mi for a range of

rock types (after Hoek & Brown, 1997).
Rock Texture
type Class Group
Coarse Medium Fine Very fine
Conglomerate Sandstone Siltstone Claystone
(22) 19 9 4
Clastic
Greywacke
(18)
Chalk
7
Organic
Coal
(8–21)
Non-
Clastic Carbonate Breccia Sparitic Micritic
(20) Limestone Limestone
(10) 8
Chemical Gypstone Anhydrite
16 13
Non-foliated Marble Hornfels Quartzite
9 (19) 24
Slightly foliated Migmatite Amphibolite Mylonites
(30) 25–31 (6)
Foliated* Gneiss Schists Phyllites Slate
33 4–8 (10) 9
Granite Rhyolite Obsidian
33 (16) (19)
Granodiorite Dacite
Light
(30) (17)
Diorite Andesite
(28) 19
Gabbro Dolerite Basalt
27 (19) (17)
Dark
Norite
22
Extrusive pyroclastic Agglomerate Breccia Tuff
type (20) (18) (15)
*
These values are for intact rock specimen tests normal to bedding or foliation. The value of mi will
be significantly different if failure occurs along a weakness plane.
where mb is the material constant for the rock mass; and s and a are constants that depend
on the characteristics of the rock mass.
The original criterion has been found to work well for most rocks of good to
reasonable quality in which the rock mass strength is controlled by tightly interlocking
angular rock pieces. The failure of such rock masses can be defined by setting a=0.5 in
Equation (4.64), giving
(4.65)
For poor quality rock masses in which the tight interlocking has been partially destroyed
by shearing or weathering, the rock mass has no tensile strength or ‘cohesion’ and
specimens will fall apart without confinement. For such rock masses the following
modified criterion is more appropriate and it is obtained by putting s=0 in Equation (4.64)
which gives
(4.66)
Equations (4.64) to (4.66) are of no practical value unless the values of the material
constants mb, s and a can be estimated in some way. Hoek and Brown (1988) proposed a
set of relations between the parameters mb, s and a and the 1976 version of Bieniawski’s
Rock Mass Rating (RMR), assuming completely dry conditions and a very favorable
(according to RMR rating system) discontinuity orientation:
(i) disturbed rock masses
(4.67a)
(4.67b)
a=0.5
(4.67c)
(ii) undisturbed or interlocking rock masses
(4.68a)
(4.68b)
a=0.5
(4.68c)
Equations (4.67) and (4.68) are acceptable for rock masses with RMR values of more
than about 25, but they do not work for very poor rock masses since the minimum value
which RMR can assume is 18 for the 1976 RMR system and 23 for the 1989 RMR
system (see Chapter 2 for details). In order to overcome this limitation, Hoek (1994) and
Hoek et al. (1995) introduced the Geological Strength Index (GSI). The relationships
between mb, s and a and the Geological Strength Index (GSI) are as follows:
(i) For GSI>25, i.e. rock masses of good to reasonable quality
(4.69a)
(4.69b)
a=0.5
(4.69c)
(ii) For GSI<25, i.e. rock masses of very poor quality
(4.70a)
s=0
(4.70b)
(4.70c)
It is noted that the distinction between disturbed and undisturbed rock masses is dropped
in evaluating the parameters mb, s and a from GSI. This is based on the fact that
disturbance is generally induced by engineering activities and should be allowed by
downgrading the values of GSI. The methods for determining RMR and GSI have been
discussed in Chapter 2.
Since many of the numerical models and limit equilibrium analyses used in rock
mechanics are expressed in terms of the Coulomb failure criterion, it is necessary to
estimate an equivalent set of cohesion and friction parameters for given Hoek-Brown
values. This can be done using a solution published by Balmer (1952) in which the
normal and shear stresses are expressed in terms of the corresponding principal stresses
as follows:
(4.71)
(4.72)
For the GSI>25, when a=0.5:
(4.73)
For the GSI<25, when s=0:
(4.74)
Once a set of (σ′n, τ) values have been calculated from Equations (4.71) and (4.72),
average cohesion c and friction angle values can be calculated by linear regression
analysis, in which the best fitting straight line is calculated for the range of (σ′n, τ) pairs.
The uniaxial compressive strength of a rock mass defined by a cohesive strength c and a
friction angle is given by
(4.75)
Water has a great effect on the strength of rock masses. Many rocks show a significant
strength decrease with increasing moisture content. Typically, strength losses of 30–
100% occur in many rocks as a result of chemical deterioration of the cement or clay
binder. Therefore, it is important to conduct laboratory tests at moisture contents which
are as close as possible to those which occur in the field. A more important effect of
water is the strength reduction which occurs as a result of water pressures in the pore
spaces in the rock. This is why the effective not the total stresses are used in the Hoek-
Brown strength criterion.
The Hoek-Brown strength criterion was originally developed for intact rock and then
extended to rock masses. The process used by Hoek and Brown in deriving their strength
criterion for intact rock (Equation 4.63) was one of pure trial and error (Hoek et al.,
1995). Apart from the conceptual starting point provided by the Griffith theory, there is
no fundamental relationship between the empirical constants included in the criterion and
any physical characteristics of the rock. The justification for choosing this particular
criterion (Equation 4.63) over the numerous alternatives lies in the adequacy of its
predictions of the observed rock fracture behavior, and the convenience of its application
to a range of typical engineering problems (Hoek, 1983). The material constant mi is

derived based upon analyses of published triaxial test results on intact rock (Hoek, 1983;
Doruk, 1991; Hoek et al., 1992). The strength criterion for rock masses is just an
empirical extension of the criterion for intact rock. Since it is practically impossible to
determine the material constants mb and s using triaxial tests on rock masses, empirical
relations are suggested to estimate these constants from RMR or GSI. The RMR and the
GSI rating systems are also empirical. For these reasons the Hoek-Brown empirical rock
mass strength criterion must be used with extreme care. In discussing the limitations in
the use of their strength criterion, Hoek and Brown (1988) emphasize that it is not
applicable to anisotropic rocks nor to elements of rock masses that behave anisotropically
by virtue of containing only a few discontinuities. Alternative empirical approaches and
further developments of the Hoek-Brown criterion which seek to account for some of its
limitations are given by Amadei (1988), Pan and Hudson (1988), Ramamurthy and Arora
(1991), Amadei and Savage (1993), and Ramamurthy (1993).
(b) Bieniawski-Yudhbir criterion

Bieniawski (1974) proposed a strength criterion for intact rock as follows
(4.76)
This was changed by Yudhbir et al. (1983), based on tests on jointed gypsum-celite
specimens, to the form
(4.77)
to fit rock masses. Yudhbir et al. (1983) recommended that the parameters α and a be
determined from
α=0.65
(4.78a)
(4.78b)
where Q is the classification index of Barton et al. (1974) and RMR is Bieniawski’s 1976
Rock Mass Rating (Bieniawski, 1976). Parameter b is determined from Table 4.6.
Kalamaras and Bieniawski (1993) suggested that both a and b should be varied with
RMR for better results. They proposed the criterion of Table 4.7 specifically for coal
seams.
(c) Johnston criterion

Based on experimental data of a wide range of geotechnical material, from lightly
overconsolidated clays through hard rocks, Johnston (1985) proposes the following
strength criterion
(4.79)
where σ′1n and σ′3n are the normalized effective principal stresses at failure, obtained by
dividing the effective principal stresses, σ′1 and σ′3, by the relevant uniaxial compressive
strength, σc; B and M are intact material constants; and s is a constant to account for the
strength of discontinuous soil and rock masses in a manner similar to that proposed by
Hoek and Brown (1980). However, in the development of the criterion, Johnston (1985)
considers only intact materials.
Table 4.6 Parameter b in the Bieniawski-Yudhbir
criterion (Yudhbir et al., 1983).
Rock Type b
Tuff, Shale, Limestone 2
Siltstone, mudstone 3
Quartzite, Sandstone, Dolerite 4
Norite, Granite, Quartz diorite, Chert 5
Table 4.7 Rock mass criterion for coal seams by

Kalamaras and Bieniawski (1993).
Equation Parameters
For intact material, s=1, the criterion becomes
(4.80)
By placing σ′3n=0, the uniaxial compressive strength is correctly modeled with the
righthand side of Equation (4.80) becoming unity.
By putting B=1, the criterion simplifies to
(4.81)
which for
(4.82)
is identical to the normalized Coulomb criterion.

The parameter B, which describes the nonlinearity of a failure envelope, is essentially
independent of the material type, and is a function of uniaxial compressive strength:
B=1−0.0172(logσc)2
(4.83)
The parameter M, which describes the slope of a failure envelope at σ′3n=0, is found to be
a function of both the uniaxial compressive strength and the material type. For the
material types shown in Table 4.8, M can be estimated by (no result is obtained for type
D material because of lack of data):
Type A, M=2.065+0.170(logσc)2
(4.84a)
2
Type B, M=2.065+0.231(logσc)
(4.84b)
Type C, M=2.065+0.270(logσc)2
(4.84c)
2
Type E, M=2.065+0.659(logσc)
(4.84d)
Table 4.8 A range of rock types (after Hoek &

Brown, 1980).
Type General Rock Type Examples
A Carbonate rocks with well developed crystal Dolomite, limestone, marble
cleavage
B Lithified argillaceous rocks Mudstone, siltstone, shale, slate
C Arenaceous rocks with strong crystals and Sandstone, quartzite
poorlydeveloped crystal cleavage
D Fine grained polyminerallic igneous crystalline Andesite, dolerite, diabase, rhyolite
rocks
E Coarse grained polyminerallic igneous and Amphibolite, gabbro, gneiss, granite,
metamorphic crystalline rocks norite, quartz diorite
(d) Ramamurthy criterion

Ramamurthy and his coworkers (Ramamurthy et al., 1985; Ramamurthy, 1986;
Ramamurthy, 1993) modified the Coulomb theory to represent the nonlinear shear
strength behavior of rocks.
For intact rock, the strength criterion is in the following form
(4.85)
where σ′1 and σ′3 are the major and minor principal effective stresses; σc is the uniaxial
compressive strength; αr is the slope of the curve between (σ′1−σ′3)/σ′3 and σc/σ′3, for
most intact rocks the mean value of αr is 0.8; and Br is a material constant of intact rock,
equal to (σ′1−σ′3)/σ′3 when σc/σ′3=1. The values of Br vary from 1.8 to 3.0 depending on
the type of rock (Table 4.9).
The values of αr and Br can be estimated by conducting a minimum of two triaxial
tests at confining pressures greater than 5% of σc for the rock. The above expression is
applicable in the ductile region and in most of the brittle region. It underestimates the
strength when σ′3 is less than 5% of σc and also ignores the tensile strength of the rock.
To account for the tensile strength, the following expression gives a better prediction for
intact rock
(4.86)
where σt is the tensile strength of rock preferably obtained from Brazilian tests; α=0.67
for most rocks; and B is a material constant. The values of α and B in Equation (4.86) can
be obtained by two triaxial tests conducted at convenient confining pressures greater than
5% of σc for the rock. In the absence of these tests, the value of B is estimated as
1.3(σc/σt)1/3.
For rock masses, the strength criterion has the same form as for intact rock, i.e.
(4.87)
Table 4.9 Mean values of parameter Br for different

rocks (after Ramamurthy, 1993).
Rock Metamorphic and sedimentary rocks
type
Argillaceous Arenaceous Chemical Igneous rocks
Siltstone Shales Sandstone Quartzite Limestone Marble Andesite Granite
Clays Slates Anhydrite Dolomite Diorite Charnockite
Tuffs Mudstone Rock salt Norite
Loess Claystone Liparite
Basalt
Br 1.8 2.2 2.2 2.6 2.4 2.8 2.6 3.0
Mean 2.0 2.4 2.6 2.8
value
where σcm is the rock mass strength in unconfined compression; Bm is a material constant
for rock mass; and αm is the slope of the plot between (σ′1−σ′3)/σ′3 and σcm/σ′3, which can
be assumed to be 0.8 for rock masses as well. σcm and Bm can be obtained by
(4.88)
(4.89)
in which σc is the unconfined compressive strength of intact rock strength; and Br is a

material constant for intact rock, as in Equation (4.83).
(e) Comments
In addition to the four empirical strength criteria for rock masses described above, there
are many other criteria. All these criteria are purely empirical and thus it is impossible to
say which one is correct or which one is not. However, the Hoek-Brown strength
criterion is the most representative one of the empirical strength criteria for rock masses,
because it is the mostly widely referred and used. Since its advent in 1980, considerable
application experience has been gained by its authors as well as by others. As a result,
this criterion has been modified several times to meet the needs of users who have
applied it to conditions which were not visualized when it was originally developed.
It is noted that all the empirical strength criteria for rock masses have the following
limitations:
1. The influence of the intermediate principal stress, which in some cases is important, is
not considered.
2. The criteria are not applicable to anisotropic rock masses. So they can be used only
when the rock masses are approximately isotropic, i.e. when the discontinuity
orientation does not have a dominant effect on failure.
4.4.2 Equivalent continuum approach for estimating rock mass strength
(a) Model of Jaeger (1960) and Jaeger and Cook (1979)

Figure 4.42(a) shows a cylindrical rock mass specimen subjected to an axial major
principal stress σ′1 and a lateral minor principal stress σ′3. The rock mass is cut by well-
defined parallel discontinuities inclined at an angle β to the major principal stress σ′1. The
strength of both the intact rock and the discontinuities are described by the Coulomb
criterion, i.e.
(4.90)
(4.91)
where τr and τj are respectively the shear strength of the intact rock and the
discontinuities; cr and are respectively the cohesion and internal friction angle of the
intact rock; cj and are respectively the cohesion and internal friction angle of the
discontinuities; and σ′n is the effective normal stress on the shear plane.
For the applied stresses on the rock mass cylinder, the effective normal stress σ′n and
the shear stress τ on a plane which makes an angle β′ to the σ′1 axis are respectively given
by
(4.92)
(4.93)
If shear failure occurs on the discontinuity plane, the effective normal stress σ′n and the
shear stress τ on the discontinuity plane can be obtained by replacing β′ in Equations
(4.92) and (4.93) by β. Adopting the obtained stresses on the discontinuity plane to
substitute for σ′n and τj in Equation (4.91) and then rearranging, we can obtain the
effective major principal stress required to cause shear failure along the discontinuity as
follows
(4.94)
If shear failure occurs in the intact rock, the minimum effective major principal stress can
be obtained by
(4.95)
The model of Jaeger (1960) and Jaeger and Cook (1979) assumes that failure during
compressive loading of a rock mass cylinder subject to a lateral stress σ′3 [see Fig.
4.42(a)] will occur when σ′1 exceeds the smaller of the σ′1f values given by Equations
(4.94) and (4.95). Figure 4.42(b) shows the variation of σ′1f with β, from which we can
clearly see the anisotropy of the rock mass strength caused by the discontinuities.
(b) Model of Amadei (1988) and Amadei and Savage (1989, 1993)
As seen above, the model of Jaeger (1960) and Jaeger and Cook (1979) assumes that the
jointed rock mass is under axisymmetric loading, so the effect of the intermediate
principal stress is not involved in their formulations. To address the limitation of the
model of Jaeger (1960) and Jaeger and Cook (1979), Amadei (1988) and Amadei and
Savage (1989, 1993) derived solutions for the strength of a jointed rock mass under a
variety of multiaxial states of stress. As in the model of Jaeger (1960) and Jaeger and
Cook (1979), the modeled rock mass is cut by a single discontinuity set. In the
formulations of Amadei (1988) and Amadei and Savage (1989, 1993), however, the
intact rock strength is described by the HoekBrown strength criterion and the
discontinuity strength is modeled using a Coulomb criterion with a zero tensile strength
cut-off.
The principle used by Amadei (1988) and Amadei and Savage (1989, 1993) to derive
the expressions of the jointed rock mass strength is the same as that used by Jaeger
(1960) and Jaeger and Cook (1979). However, since the effect of the intermediate
principal stress is included and since the nonlinear Hoek-Brown strength criterion is used,
the derivation process and the final results are much more complicated. For reasons of
space, only some of the typical results of Amadei and Savage (1989, 1993) are shown
here.
Consider a jointed rock mass cube under a triaxial state of stress σ′x, σ′y and σ′z. The
orientation of the discontinuity plane is defined by two angles β and Ψ with respect to the
xyz coordinate system (see Fig. 4.43). Let nst be another coordinate system attached to
the discontinuity plane such that the n-axis is along the discontinuity upward normal and
the s-and t-axes are in the discontinuity plane. The t-axis is in the xz plane. The upward
unit vector n has direction cosines
Fig. 4.42 Variation of compressive

strength with angle β of the
discontinuity plane (after Jaeger &
Cook, 1979).
(4.96)
Defining m=σ′y/σ′x and n=σ′z/σ′x and introducing two functions
(4.97)
where σ′n and τ are respectively the normal and shear stresses acting across the
discontinuity; and is the friction angle of the discontinuity, the limiting equilibrium
(incipient slip) condition of the discontinuity can be derived as
(4.98)
The nonnegative normal stress condition of the discontinuity is
(4.99)
So for a discontinuity with orientation angles β and ψ the condition Ff=0 corresponds to
impending slip. No slip takes place when Ff is negative. Figure 4.44 shows a typical set
of failure surfaces Ff(m,n)=Q for ψ equal to 40° or 80° and β ranging between 0° and 90°.
In this figure the ranges Ff(m,n)>0 are shaded and Fn=0 is represented as a dashed
straight line. The positive normal stress condition (Fn>0) is shown as the region on either
side of the line Fn=0 depending on the sign of σx.
Fig. 4.43 Discontinuity plane in a

triaxial stress field (after Amadei &
Savage, 1993).
Depending on the ordering of σ′x, σ′y and σ′z, the Hoek-Brown strength criterion for intact
rock (Equation 4.63) assume six possible forms as shown in Table 4.10. Using mi=
Table 4.10 Forms of Equation (4.63) for different

orderings of σ′x, σ′y and σ′z.
Principal stress Major stress Minor stress Forms of Equation (4.63)
ordering σ′1 σ′3
σ′x>σ′y>σ′z σ′x σ′z
σ′x>σ′z>σ′y σ′x σ′y
σ′y>σ′x>σ′z σ′y σ′z
σ′y>σ′z>σ′x σ′y σ′x
σ′z>σ′x>σ′y σ′z σ′y
σ′z>σ′y>σ′x σ′z σ′x

Fig. 4.44 Shape of the failure surface

Ff(m,n)=0 in the m=σ′y/σ′x, n=σ′z/σ′x
space for (a) β= 38.935°, ψ=40°; (b)
β=30°, ψ=40°; (c) β=20°, ψ=80°; and
(d) β=70°, ψ=40°. The region Fn >0 is
above the dashed line Fn=0 when σ′x is
compressive and below that line when
σ′x is tensile. Friction (after
Amadei & Savage, 1993).
7 and σc=42 MPa, the intact rock failure surfaces for different values of σ′x/σc can be
obtained as shown in Figure 4.45.
The failure surfaces of the jointed rock masses can be obtained by superposition of the
discontinuity failure surfaces and the intact rock failure surfaces. Figure 4.46 is obtained
by superposition of the failure surfaces in Figures 4.44 and 4.45. The following remarks
can be made about the diagrams shown in Figure 4.46:
1. In general, for a given value of σ′x/σc, the size of the stable domain enclosed by the
intact rock failure surface is reduced because of the discontinuities. The symmetry of
the intact rock failure surface with respect to the m=n axis in the m, n space (Fig. 4.
46) is lost. The strength of the jointed rock mass is clearly anisotropic.
2. The strength reduction associated with the discontinuities is more pronounced for
discontinuities with orientation angles β and ψ for which the discontinuity failure
surface in the m, n space is ellipse than when it is an hyperbola or a parabola.
3. Despite the zero discontinuity tensile strength and the strength reduction associated
with the discontinuities, jointed rock masses can be stable under a wide variety of
states of stress σ′x, σ′y=mσ′x, σ′z=nσ′x. These states of stress depend on the values of
discontinuity orientation angles β and ψ and the stress ratio σ′x/σc.
(c) Comments
In Section (a), a rock mass with one discontinuity set is considered. If we apply the model
of Jaeger (1960) and Jaeger and Cook (1979) to a rock mass with several discontinuity
sets, the strength of the rock mass can be obtained by considering the effect of each
discontinuity set. For example, consider a simple case of two discontinuity sets A and B
[see Fig. 4.47(a)], the angle between them being α. The corresponding variation of the
compressive strength σ′1β, if the two discontinuity sets are present singly, is shown in
Figure 4.47(b). As the angle βa of discontinuity set A is changed from 0 to 90°, the angle
βb of discontinuity set B with the major stress direction will be
βb=|α−βa| for α≤90°
(4.100)
When βa is varied from 0 to 90°, the resultant strength variation for α=60 and 90° will be
as in Figure 4.47(c), choosing the minimum of the two values σ′1βa and the corresponding
σ′1βb from the curves in Figure 4.47(b).
Hoek and Brown (1980) have shown that with three or more discontinuity sets, all sets
having identical strength characteristics, the rock mass will exhibit an almost flat strength
variation (see Fig. 4.48), concluding that in highly jointed rock masses, it is possible to
adopt one of the rock mass failure criteria presented in Section 4.4.1.
It should be noted that, in the models of the equivalent continuum approach,
discontinuities are assumed to be persistent and all discontinuities in one set have the
same orientation. In reality, however, discontinuities are usually non-persistent and the
discontinuities in one set have orientation distributions.
Fig. 4.45 Geometrical representation of

the Hoek-Brown failure surface for
intact rock in the m= σ′y/σ′x, n=σ′z/σ′x
space for different values of σ′x/σc with
mi=7 and σc=42 MPa. (a) σ′x is
compressive; and (b) σ′x is tensile
(after Amadei & Savage, 1993).
Fig. 4.46 Superposition of the joint

failure surface with and the
intact rock failure surface with mi=7
and σc=42 MPa in the m=σ′y/σ′x,
n=σ′z/σ′x space for (a) β=38.935°,
ψ=40°; (b) β= 30°, ψ=40°; (c) β=20°,
ψ=80°; and (d) β=70°, ψ=40° (after
Amadei & Savage, 1993).
4.4.3 Direct consideration of discontinuities in numerical analysis

The effect of discontinuities on the rock mass strength can be directly accounted for in
numerical analyses. Special joint elements used to represent the discontinuities in FEM
analysis have been discussed in Section 4.3.3. Here only the constitutive models of
discontinuities will be described and discussed.
To account for the effect of discontinuities on the rock mass strength in numerical
analysis, the behavior of discontinuities is usually assumed to be elasto-plastic. The
elastic behavior is represented by the initial elastic tangential normal and shear stiffnesses
kne and kse (see Fig. 4.1). The peak strength and dilatancy of rock discontinuities is
represented by a failure criterion and a flow rule, respectively.
(a) Coulomb Model

The Coulomb model is perhaps the crudest of rock discontinuity models but has been
extensively used in engineering analysis and design of rock structures. The Coulomb
strength criterion has been given in Equation (4.8) and is rewritten here in the following
form
(4.101)
where |σs| is the absolute value of shear stress on the discontinuity plane; cj and are
respectively the cohesion and friction angle of the discontinuities; and σ′n is the effective
normal stress on the discontinuity plane.
If an associated flow rule is adopted, rates of plastic normal strain and shear strain
are given by
(4.102)
where is a positive proportionality constant. Equation (4.102) implies
(4.103)
Fig. 4.47 (a) Rock mass with two

discontinuity sets A and B; (b)
Strength variation with β if the
discontinuity sets are present singly;
and (c) Strength variation when both
discontinuity sets are present.
Fig. 4.48 Strength variation with angle

β1 of discontinuity plane 1 in the
presence of 4 discontinuity sets, the
angle between two adjoining planes
being 45° (after Hoek & Brown, 1980).
The discontinuities are, therefore, dilatant, i.e. an increment of shear displacement ∆us
along the discontinuity is accompanied by an increment in the normal displacement ∆un
given by
(4.104)
The rate of dilation is constant and goes on unabated. This behavior is quite unrealistic.
Roberds and Einstein (1978) presented a very comprehensive model for rock
discontinuities. From various studies it has been established that the flow rule for rock
discontinuities should be non-associated. By introducing a variable dilation angle Ψj, a
plastic potential function can be written as
(4.105)
where Ψj can be identified from the experimental results on rock discontinuities. It is

clear that when the average normal displacement on the rock discontinuity is equal to the
average height of the asperities, dilation must cease, i.e. Ψj→0.
The Coulomb model has another drawback. cj and in Equation (4.101) are not truly
constants. They depend on σ′n. The values of σ′n on rock discontinuities can vary by
several orders of magnitudes within the structure to be analyzed. Choosing a single
appropriate value of cj and for a discontinuity set, therefore, becomes difficult, if not
impossible.
(b) Barton Model

The Barton model has been described in Section 4.2.2 [Equation (4.12)] and is rewritten
here in the following form
(4.106)
where JRC is the discontinuity roughness coefficient; JCS is the discontinuity wall
compressive strength; and is the basic friction angle of the rock material.
If an associated flow rule is assumed, the dilation angle at peak strength can be readily
computed by differentiating Equation (4.106). However, the computed dilation angles Ψj
based on an associated flow rule do not match the experimentally observed values. This
again shows that the flow rule for rock discontinuities should be non-associated. Pande
and Xiong (1982) proposed the following plastic potential function to match the
experimental results of Barton and Chaubey (1977):
(4.107)
where
(4.108)
Table 4.11 shows the comparison of experimental values with those computed using
Equation (4.107) as the plastic potential function. A close agreement can be seen.
(c) Comments
In addition to the two elasto-plastic models for rock discontinuities described above,
there are many other models. Roberds and Einstein (1978) presented a very
comprehensive model and critically examined Patton (1966) model, Ladanyi and
Archambault (1970) model, Agbabian model (Ghaboussi et al., 1973), Goodman (1966,
1974) model and Barton (1976) model by comparing them with the comprehensive
model. Since the comprehensive rock discontinuity model of Roberds and Einstein
(1978) can treat the entire behavioral history from the creation of the discontinuity to its
behavior before, during and after sliding, it provides a good basis for comparison of
various models. With the comprehensive rock discontinuity model, it is possible to show
where and to what extent the existing models are limited or simplified as compared to the
comprehensive model and this makes it possible to appropriately modify the existing
models, if so desired.
Table 4.11 Comparison of measured angle of
dilation with that predicted by Equation (4.107).
Rock Type No. of Measured angle of dilation Computed angle of dilation
Samples (°) (°)
Alpite 36 25.5 23.0
Granite 38 20.9 20.2
Hornfels 17 26.5 26.2
Calcareous 11 14.8 19.1
shale
Slate 7 6.8 –
Gneiss 17 17.3 15.5
Soapstone 5 16.2 18.6
model
Fractures 130 13.2 –
4.5 SCALE EFFECT
Research results (see, e.g., Heuze, 1980; Hoek & Brown, 1980; Medhurst & Brown,
1996) indicate that rock masses show strong scale dependent mechanical properties. In
the following, the scale effect on the strength and deformation properties of rock masses
is briefly discussed.
4.5.1 Scale effect on strength of rock mass

Experimental results show that rock strength decreases significantly with increasing
sample size. Based upon an analysis of published data, Hoek and Brown (1980)
suggested that the unconfined compressive strength σcd of a rock specimen with a
diameter of d mm is related to the unconfined compressive strength σc50 of a 50 mm
diameter specimen by
(4.109)
This relationship, together with the data upon which it was based, is illustrated in Figure
4.49. Hoek and Brown (1997) suggested that the reduction in strength is due to the
greater opportunity for failure through and around grains, the “building blocks” of intact
rock, as more and more of these grains are included in the test sample. Eventually, when
a sufficiently large number of grains are included in the sample, the strength reaches a
constant value.
Medhurst and Brown (1996) reported the results of laboratory triaxial tests on 61, 101,
146 and 300 mm diameter samples of coal from the Moura mine in Australia. The results
of these tests are as summarized in Table 4.12 and Figure 4.50. It can be seen that the
strength decreases significantly with increasing specimen size. This is attributed to the
effects of cleat spacing. For this coal, the persistent cleats are spaced at 0.3–1.0 m while
non-persistent cleats within vitrain bands and individual lithotypes define blocks of 1 cm
or less. This cleating results in a “critical” sample size of about 1m above which the
strength remains constant.
Heuze (1980) conducted an extensive literature search and found results of 77 plate tests
as shown in Figure 4.51. The test volume shown in this figure is calculated in the
following way:
1. For a circular plate, the test volume is taken as that of a sphere having a diameter of 4
times the diameter of the plate.
2. For a rectangular or square plate of given area, the diameter of a circle of equal area is
first calculated, and the test volume is then determined using the equivalent diameter.
Fig. 4.49 Influence of specimen size on

the strength of intact rock (after Hoek
& Brown, 1980).
Table 4.12 Peak strength of Moura coal in terms of
the parameters in Equation (4.64), based upon a
value of σc=32.7 Mpa.
Diameter (mm) mb s a
61 19.4 1.0 0.5
101 13.3 0.555 0.5
146 10.0 0.236 0.5
300 5.7 0.184 0.6
Mass 2.6 0.052 0.65
Fig. 4.50 Peak strength for Australian

Moura coal (after Medhurst & Brown,
1996).
The number shown next to the open triangles in the figure indicates the number of tests
performed; the mean value of these test results is plotted as the triangle. The test results
[except those of Coates and Gyenge (1966) and Rhodes (1973)] show that the strength
decreases with increasing test volume.
Figure 4.52 (Hoek et al., 1995) shows a simplified representation of the influence of the
relation between the discontinuity spacing and the size of the problem domain on the
selection of a rock mass behavior model (Hoek-Brown strength criterion). As the

problem domain enlarges, the corresponding rock behavior changes from that of the
isotropic intact rock, through that of a highly anisotropic rock mass in which failure is
controlled by one or two discontinuities, to that an isotropic heavily jointed rock mass.
In determining the allowable bearing capacity of shallow foundations on rock using
the Hoek-Brown strength criterion, Serrano and Olalla (1996), following the technique of
Hoek (1983), divide the rock masses into three main structural groups depending on the
conditions of rock masses and the foundation dimensions (Fig. 4.53):
Fig. 4.51 Effect of test volume on

measured bearing strength of rock
masses. The number next to the
triangle indicates the number of tests
performed (after Heuze, 1980).
Fig. 4.52 Simplified representation of

the influence of scale on the type of
rock mass behavior (after Hoeketal,
1995).
Group I: where the rock can be considered as intact. If the microstructure

of the rock is isotropic, the rock mass can be considered isotropic and the
Hoek-Brown criterion can be applied.
Group II: where the rock mass is affected by only a few sets of
discontinuities. The behavior of the rock mass is basically anisotropic and
the Hoek-Brown criterion cannot generally be applied to the rock mass.

scale effect on the type of rock mass
model which should be used in
designing shallow foundations on rock
slope (after Serrano & Olalla, 1996).
Group III: where the rock mass is affected by a number of sets of

discontinuities giving rise to “small spacing” between discontinuities.
This group of rock masses can be regarded as isotropically broken media
and the Hoek-Brown criterion can be applied.
“Small spacing” is a relative concept, in the sense that it depends on the foundation
dimensions. Serrano and Olalla (1996) propose a parameter, the “spacing ratio of a
foundation” (SR), for its quantification. SR is defined as
(4.110)
where B is the foundation width (in meters); smi is the discontinuity spacing of set i (in
meters); λi is the frequency of discontinuity set i (m−1); and n is the number of
discontinuity sets.
As an initial and conservative proposal, a “relatively small spacing” is suggested when
SR is greater than 60. A value of 60 means that, if there are four sets of discontinuities,
each of them appears at least 15 times within the foundation width. When SR>60, the
mass can be regarded as an isotropically broken medium and the Hoek-Brown criterion
can be applied.
For values of SR≤(0.8−4), in the case of four sets of discontinuities, the rock mass can
be considered as an intact rock mass (Group I).
4.5.2 Scale effect on deformability of rock mass

The scale effect on the deformability of rock masses can be seen from the difference of
rock mass modulus measured in the field and intact rock mass modulus measured in the
laboratory. Heuze (1980) concluded that the rock mass modulus measured in the field
ranges between 20 and 60% of the intact rock mass modulus measured in the laboratory.
One simple and apparent explanation to the reduction of rock mass modulus is that the
effect of discontinuities is included in the rock masses.
4.6 DISCUSSION
The structure of jointed rock masses is highly variable; the methods used to consider the
effect of discontinuities on the mechanical behavior of jointed rock masses are also
variable. The selection of the methods should be based on careful studies of the in situ
situation of jointed rock masses.
Laboratory and in situ tests (i.e., direct methods) can directly provide results about the
mechanical properties of tested specimens. However, care need be exercised about the
extent to which the measured behavior of the rock specimen reflects the actual behavior
of rock masses. The extrapolation of the behavior induced by the experimental system to
different circumstances can be very misleading. In addition, in situ tests are time
consuming, expensive and difficult to conduct; it is extremely difficult to investigate the
effects of discontinuity system on the mechanical properties of jointed rock masses
through in situ tests.
Indirect methods consist of the empirical methods, the equivalent continuum approach
and numerical analysis methods. It is important to note that all the indirect methods need
to use some of the mechanical properties of intact rock or discontinuities obtained
through laboratory or in situ tests.
Since they are simple and easy to use, and most importantly, since they originate from
practical experience, the empirical methods are most widely used in design practice.
However, it is important to note their limitations as described in Sections 4.3.1 and 4.4.1.
The equivalent continuum approach usually assumes that all discontinuities are
persistent and the discontinuities in one set have the same orientation. In reality, however,
discontinuities are usually non-persistent and the discontinuities in one set are not in the
same orientation. Kulatilake et al. (1992, 1993) and Wang (1992) considered rock masses
containing non-persistent discontinuities and derived relationships between the

deformation properties and the fracture tensor parameters from the DEM analysis results
of generated rock mass blocks. However, this method is also limited as described in
Section 4.3.2.
The numerical methods have great potential for the complex mechanical analyses of
jointed rock masses. The key problems associated with numerical methods are the
representation of discontinuities and the determination of discontinuity constitutive
models. The main drawback of this approach is that, due to computer limitations and
difficulty in creating meshes for a heavily jointed rock mass, only rock masses with a
limited number of discontinuities can be analyzed.
In summary, the limitations for each method are as follows:
1. For laboratory tests, only small specimens can be used. Since rock masses show strong
scale dependent mechanical properties, the measured behavior of small rock
specimens may not reflect the actual behavior of rock masses in the field.
2. In situ tests are time consuming, expensive and difficult to conduct.
3. Empirical methods do not consider the anisotropy of rock masses caused by
discontinuities and different empirical relations often give very different values.
4. The equivalent continuum approach assumes that the discontinuities are persistent and
the discontinuities in one set have the same orientation. In reality, however, most of
the discontinuities are non-persistent with finite size and the discontinuities in one set
are not in the same orientation.
5. Numerical methods can only be used for rock masses with a limited number of
discontinuities.
Because each method has its own advantages and disadvantages, it is important to select
the appropriate method(s) for different purposes. Following are the principles that can be
used when selecting the methods according to the nature of the problems:
1. The empirical methods can always be used in the first stage of design. For heavily
jointed rock masses, the empirical methods can be used in all design stages.
2. For rock masses with persistent discontinuities which are regularly distributed, the
equivalent continuum approach can be used.
3. For rock masses with a limited number of discontinuities, the numerical methods and
the limit equilibrium method can be used.
5
Site investigation and rock testing
5.1 INTRODUCTION
As required for any geotechnical projects, site investigations need be conducted to obtain
the information required for the design of drilled shafts in rock. The nature and extent of
the information to be obtained from a site investigation will vary according to the project
involved and the expected ground conditions. A site investigation is a process of
progressive discovery, and, although there must be a plan and program of work at the
beginning, the information emerging at any stage will influence the requirements of
subsequent stages. Typically, a site investigation consists of the following three main
stages:
1. Preliminary investigation including desk study and site reconnaissance
2. Detailed investigation including boring, drilling, in situ testing and lab testing
3. Review during construction and monitoring
A distinguishing feature of site investigations for foundations in rock is that it is
particularly important to focus on the details of the structural geology. The rock mass at a
site may contain very strong intact rock, but the discontinuities in the rock mass may lead
to excessive deformation or even failure of the drilled shaft foundations in the rock mass.
5.1.1 Preliminary investigation

Prior to implementing a detailed site investigation program, certain types of preliminary
information need be developed. The type and extent of information depends on the cost
and complexity of the project. The information is developed from a thorough survey of
existing information and field reconnaissance. Information on topography, geology and
potential geologic hazards, surface and groundwater hydrology, seismology, and rock
mass characteristics are reviewed to determine the following (ASCE, 1996):
• Adequacy of available data
• Type and extent of additional data that will be needed
• The need for initiating critical long-term studies, such as ground water and seismicity
studies, that require advance planning and early action
• Possible locations and type of geologic features that might control the design and
construction of drilled shaft foundations.
Various types of published maps can provide an excellent source of geologic information
to develop the regional geology and geological models of potential or final sites. Other
geotechnical information and data pertinent to the project can frequently be obtained
from a careful search of federal, state, or local governments as well as private industry in
the vicinity. Consultation with private geotechnical engineering firms, mining companies,
well drilling and development companies and state and private university staff can
sometimes provide a wealth of information.
After a complete review of available geotechnical data, a site reconnaissance should
be made to gather information through visual examination of the site and an inspection of
ground exposures in the vicinity. In some cases adjacent sites will also be examined. The
primary objective of this field reconnaissance is to, insofar as possible, confirm, correct
or expand geologic and hydrologic information collected from preliminary office studies.
If rock outcrops are present, the field reconnaissance offers an opportunity to collect
preliminary information on rock mass conditions that might influence the design and
construction of drilled shafts. Notation should be made of the strike and dip of major
discontinuity sets, discontinuity spacing, discontinuity conditions (i.e. weathering, wall
roughness, tightness, fillings, and shear zones), and discontinuity persistence.
The reconnaissance will assist in planning the detailed investigation program. Where
the geology is relatively straightforward and the engineering problems are not complex,
sufficient geological information may be provided by the desk study, subject to
confirmation by the exploration which follows. In other cases detailed investigations may
be carried out.
Geophysical survey is becoming quicker and more robust to provide information on
the depth of weathering, the bedrock profile, the location of major faults and solution
cavities, and the degree of fracturing of the rock. So some geophysical work is often
conducted in the stage of preliminary investigation rather than leaving it all to the stage
of detailed investigation. In some cases it may be appropriate to put down pits or use
relatively light and simple boring equipment during the preliminary investigation.
However, the objectives of a boring program at this stage should be limited. The main
boring program should be deferred until the stage of detailed investigation.
5.1.2 Detailed investigation

Information from the desk study and site reconnaissance provides a preliminary
conception of the ground conditions and the engineering problems that may be involved.
Detailed investigation then proceeds with flexible planning so that the work can be varied
as necessary as fresh information emerges. The extent of the detailed investigation will be
governed by the type of project and the nature and variation of the ground. Other factors
to be taken into account are access, time available and cost. However, technical
requirements rather than cost should govern.
Various methods are available for detailed investigation of rock masses. This chapter
will describe discontinuity sampling on exposed rock faces (Section 5.2), boring (Section
5.3), geophysical exploration (Section 5.4), lab testing (Section 5.5) and in situ testing
(Section 5.6).
5.7.3 Review during construction and monitoring

It is essential to examine all excavations during construction to see whether the
expectations of the preceding investigations have been realized. The examinations can be
carried out after the excavation has been cleaned up and just prior to the placement of
Site investigation and rock testing 175
concrete. The identification of exceptions may lead to an early diagnosis and anticipation
of problems.
During construction and in the post-commissioning stage, monitoring will involve
regular reading of instruments installed to check performance against design criteria. This
should serve as an “early warning” system, which will initiate a contingency program,
thus minimizing the delays that would occur as a result of an adverse situation.
5.2 DISCONTINUITY SAMPLING ON EXPOSED ROCK FACES
From sampling on exposed rock faces, either above or below ground, information about
the orientation, spacing, roughness and curvature of discontinuities can usually be
satisfactorily obtained. It should be noted, however, that there exists sampling bias on
discontinuity orientation and spacing (Terzaghi, 1965; Priest, 1993; Mauldon &
Mauldon, 1997). This sampling bias should be corrected before inferring statistical
distributions of orientation data. Although the locations of traces give some information
about the locations of discontinuities, it is still impossible to determine the exact locations
of discontinuities. From sampling on exposed rock faces, almost no information about the
shape of discontinuities can be obtained. Measured trace lengths give some information
about the size of discontinuities. However, because of the sampling biases and the
unknown shape of discontinuities, the size of discontinuities can only be inferred based
on assumptions (see Chapter 3).
5.2.1 Scanline sampling

Intersections between discontinuities and the rock face produce linear traces which
provide an essentially two-dimensional sample of the discontinuity network. In current
geologic and rock engineering practice, straight scanlines are commonly used for
discontinuity sampling on exposed rock faces (see Fig. 5.1). The straight scanlines can be
simply measuring tapes pinned with masonry nails and wire to the rock face or chalk
lines drawn on the rock face. The length and orientation of each discontinuity crossing
the straight scanline are measured. Spacing between adjacent discontinuities intersecting
the straight scanline can also be measured along the scanline. From measurements taken
this way, three inferences can be made on the population of discontinuities: intensity (or
spacing), size (length or persistence), and orientation.
Straight scanlines provide a fast method for discontinuity sampling, but yield a sample
biased by the relative orientation and size between the discontinuities and the rock face.
The relative orientation introduces a bias to both the discontinuity spacing and the
number of discontinuities that are measured. The bias arises because discontinuities sub-
parallel to the rock face have less chance to intersect the rock face than discontinuities
perpendicular to the rock face (Terzaghi, 1965). The bias in spacing can be corrected as
follows (Terzahgi, 1965):
s=sa sinθ
(5.1)
where s is the true spacing between discontinuities of the same set; sa is the measured
(apparent) spacing between discontinuities of the same set on the rock face; and θ is the
angle between the scanline and the discontinuity traces (see Fig. 5.1).
The number of discontinuities in a set can be adjusted to account for the orientation
bias as follows:
(5.2)
where N is the adjusted number of discontinuities; Na is the measured (apparent) number

of discontinuities.
The statistical analysis of discontinuity spacing and frequency has been discussed in
Chapter 3.
Fig. 5.1 Straight scanline sampling.

There is also sampling bias for trace lengths intersected by a straight scanline because the
scanline tends to intersect preferentially the longer traces (Priest & Hudson, 1981). For
detailed analysis of sampling biases on trace lengths, the reader can refer to Priest (1993).
Circular scanlines (see Fig. 5.2) are also used for discontinuity sampling at exposed
rock faces (Einstein et al., 1979; Titley et al., 1986; Davis & Reynolds, 1996; Mauldon et
al., 2000). Circular scanline sampling measures only the traces intersecting the line of the
circle. One advantage of circular scanlines over straight scanlines is the elimination of
directional bias. Mauldon et al. (2000) derived a simple expression for estimating
discontinuity intensity from circular scanline sampling:
(5.3)
where λ′ is the discontinuity intensity defined as the mean length of traces per unit area;
N is the number of traces intersecting the circular scanline; and c is the radius of the
scanline circle.
Fig. 5.2 Circular scanline sampling.
5.2.2 Window sampling

Window or area sampling measures discontinuity traces within a finite size area (usually
rectangular or circular in shape as discussed in Chapter 3) at an exposed rock face (see
Figs. 3.22 and 3.23). It is important to note that in window sampling as it is defined here
only the portions of the discontinuity traces within the window are measured, while the
portions of traces intersecting such a window but lying outside are not considered.
Window sampling reduces the sampling biases for orientation and size created by
scanline sampling, but problems of discontinuity curtailment remain where the rock face
is of limited extent.
The estimation of mean trace length from sampling on rectangular or circular windows
has been described in Chapter 3.
The probability of intersecting a discontinuity with a sampling domain depends on the
relative orientation of the discontinuity with respect to the sampling domain, the shape
and size of the discontinuity, and the shape and size of the sampling domain. Therefore,
observed frequencies of discontinuities do not represent the true frequencies in three
dimensions. This is called the sampling bias on discontinuity orientation. This sampling
bias should be corrected when inferring statistical distributions of orientation data.
Kulatilake and Wu (1984a) proposed a method to find the probability that a finite size
discontinuity intersects a finite size sampling plane, and presented a procedure for
correcting sampling bias for circular discontinuities intersecting rectangular, vertical
sampling planes. Kulatilake et al. (1990) extended the previous formulation to cover
other discontinuity shapes such as parallelograms, rectangles, rhombuses and triangles.
However, the procedures for correcting sampling bias cannot be applied directly for non-
vertical, finite sampling areas. Wathugala et al. (1990) presented, using a vector
approach, a more general procedure for correcting sampling bias on orientation,
applicable for sampling planes of any orientation.
5.2.3 Photographic mapping

Field sampling of discontinuities requires sufficient exposed and accessible rock faces,
time and considerable cost. Photographic techniques provide a way of overcoming the
above difficulties (Hagan, 1980; Thomas et al., 1987; Franklin et al., 1988; BlinLacroix
et al., 1990; Tsoutrelis et al., 1990; Ord & Cheung, 1991; Crosta, 1997). Application of
photographic techniques provides a partly automated method for estimating discontinuity
characteristics including orientation, size, spacing and surface geometry. Photographs of
the rock face have to get through the digitizing and processing phase before performing
any correction and analysis. Many image processing software programs exist but,
according to Crosta (1997), the best way is still the digitization of fracture lineaments (2-
D) by means of a digitizing board. This approach allows for a more correct representation
of the fractures with respect to other automatic procedures. In fact, the operator, who
preferably should be the same person who accomplished the field survey, is allowed to
select or discard lineaments that can be ascribed to or not to the discontinuity network
(vegetation, shadows, facets). The digitized photographs can then be interrogated by
computerized sampling. The sampling techniques can vary from a single scanline or
window to multiple scanlines or windows. Discontinuity characteristics obtained from
photographic mapping have been found to agree well with values obtained using standard
manual procedures (Tsoutrelis et al., 1990; Crosta, 1997).
5.3 BORINGS
Borings, in most cases, provide the only viable exploratory tool that directly reveals
geologic evidence of the subsurface site conditions. In addition to exploring geologic
stratigraphy and structure, borings are necessary to obtain samples for laboratory
engineering property tests. Borings are also frequently made for other purposes, such as
collection of groundwater data, performing in situ tests, installing instruments, and
exploring the condition of existing structures. Of the various boring methods, rock core
borings are the most useful in rock foundation investigations.
5.3.7 Rock core boring

Rock core boring is the process in which diamond or other types of core drill bits are
used to drill exploratory holes and retrieve rock cores. Good rock core retrieval with a
minimum of disturbance requires the expertise of an experienced drill crew. Core bits that
produce 2.0 inch (nominal) diameter cores (i.e., NW or NQ bit sizes) are satisfactory for
most exploration work in good rock as well as provide sufficient size samples for most
rock index tests such as unconfined compression, density, and petrographic analysis.
However, the use of larger diameter core bits ranging from 4.0 to 6.0 inches (nominal) in
diameter are frequently required to produce good cores in soft, weak and/or fractured
strata. The larger diameter cores are also more desirable for samples from which rock
strength test specimens are prepared; particularly strengths of natural discontinuities. The
number of borings and the depths to which bore holes should be advanced are dependent
upon the subsurface geological conditions, the project site areas, types of projects and
structural features. Where rock mass conditions are known to be massive and of excellent
quality, the number and depth of borings can be minimal. Where the foundation rock is
suspected to be highly variable and weak, such as karstic limestone or sedimentary rock
containing weak and compressible seams, one or more borings for each major load
bearing foundation element may be required. In cases where structural loads may cause
excessive deformation, at least one of the boreholes should be extended to a depth
equivalent to an elevation where the structure imposed stress acting within the foundation
material is no more than 10 percent of the maximum stress applied by the foundation.
While the majorities of rock core borings are drilled vertically, inclined borings and in
some cases oriented cores are required to adequately define stratification and jointing. In
near vertical bedding, inclined borings can be used to reduce the total number of borings
needed to obtain core samples of all strata. Where precise geological structure is required
from core samples, techniques involving oriented cores are sometimes employed. In these
procedures, the core is scribed or engraved with a special drilling tool so that its
orientation is preserved. In this manner, both the dip and dip orientation of any joint,
bedding plane, or other planar surface can be ascertained.
To ensure the maximum amount of data recovered from rock core borings it is
necessary to correctly orient boreholes with respect to discontinuities present in the rock
mass. If there is an outcrop present the main discontinuity sets should be established and
the borehole(s) drilled to intersect these sets at as large an angle as possible. If no outcrop
is present, the discontinuity pattern is unknown, and to ensure representative results, a
minimum of three holes should be drilled as nearly orthogonal to each other as possible
(ISRM, 1978; McMillan et al., 1996).
(a) Core logging

From core logging, one can obtain the total core recovery, discontinuity frequency, rock
quality designation (RQD) and other discontinuity information such as orientation,
spacing and aperture (ISRM, 1978). Before making detailed observations the core as a
whole should be examined to determine the structural boundaries (domains) and
geological features to be measured. The markers indicating depths of geological horizons
and the start and end of each run should be carefolly checked for errors.
Total core recovery is defined as the summed length of all pieces of recovered core
expressed as a percentage of the length drilled and should be measured and recorded to
the nearest 2% if possible. When the core is highly fragmented the length of such
portions is estimated by assembling the fragments and estimating the length of core that
the fragments appear to represent. Core recovery is normally used to describe individual
core runs or whole boreholes, and not specific structurally defined rock units. The results
obtained in a rock mass of poor quality will be strongly dependent on the drilling
equipment and the skill of the drilling crew. Core grinding may result in excessive lost
core. Core that is damaged in this way should always be recorded. The depth drilled at
start and end of zones of core loss should be carefully recorded. The relevant lengths lost
can be replaced by wooden blocks with markings on both ends.
Frequency is defined as the number of natural discontinuities intersecting a unit length
of recovered core and should be counted for each meter of core. Artificial fractures
resulting from rough handling or from drilling process should be discounted only when
they can be clearly distinguished from natural discontinuities. It should be noted that
orientation bias need be corrected in order to obtain the true discontinuity frequency. This
can be done by treating the core axis as a scanline and using Equation (5.2).
Discontinuity spacing may also be estimated by matching the individual core pieces
and measuring the length along the core axis between adjacent natural discontinuities of
one set. Again the orientation bias need be corrected in order to obtain the true
discontinuity spacing (Equation 5.1).
Terzaghi’s (1965) method (Equations 5.1 and 5.2) for correcting the orientation bias
assumes discontinuities of infinite size and does not consider the effect of borehole size.
For discontinuities of finite size intersecting a borehole, the size of both the discontinuity
and borehole will influence the probability of intersection. Mauldon and Mauldon (1997)
developed a procedure for correcting orientation bias when sampling discontinuities
using a borehole. In their approach, discontinuities are assumed to be discs of finite size
and the borehole is assumed to be an infinitely long cylinder of circular cross section.
Rock quality designation (RQD) is a modified core recovery percentage in which all
the pieces of sound core over 4 in. (10.16 cm) long are counted as recovery, and are
expressed as a percentage of the length drilled. The small pieces resulting from closer
jointing, faulting or weathering are discounted. The detailed procedure for estimating
RQD has been described in Chapter 2.
(b) Core orientation

To determine the dip and dip orientation of discontinuities from core samples, the core
orientation need be known. The following briefly describes some of the techniques for
determining the core orientation.
Craelius Method (Goodman, 1976; ISRM, 1978) Orientation of the core is based on
orienting the first piece of core in each core run. The orienting device is a cylinder, of
about the same diameter as the core, with six locking extension feet. At the beginning of
each core run, it is inserted on the front of the core barrel with the feet fully extended. It
is lowered with the core barrel and when it hits the bottom, the feet are depressed
differentially until they lock into position. The orienting device rides into the barrel as
coring progresses. When the core barrel is emptied, the top of the core is laid in an
alignment cradle against the Craelius device and rotated to find the proper fit of the feet
against the rough top-of-core, and the orientation of the first piece of core can thus be
determined. The orientation of other pieces is determined by laying them on a “V” trough
in proper sequence and rotating them as necessary to fit all pieces together. This method
works well if adjacent pieces of core can be matched. Zones of core loss and
perpendicularly intersected discontinuities reduce the effectiveness of the method locally.
Christensen-Huegel Method (Goodman, 1976; ISRM, 1978) The ChristensenHuegel

barrel contains three knives on a shoe mounted on the end of the inner barrel, so that as
the core enters the core holder, three grooves are cut longitudinally. The barrel also has a
compass photo device to give the bearing of the hole and the orientation of a marker
oriented relative to one of the scribing lines.
Integral Sampling Method (Goodman, 1976; ISRM, 1978) The cores are first reinforced
with a grouted bar whose azimuth is known from positioning rods. The reinforcing bar is
co-axially overcored with a large diameter coring crown (see Fig. 5.3).
Clay Core Barrel Method (Call et al., 1981) A modified inner core barrel is used with
conventional diamond drilling equipment (see Fig. 5.4). The barrel is eccentrically
weighted with lead and lowered into an inclined, fluid-filled borehole so that its
orientation with respect to the vertical is known. Modeling clay protrudes from the
downhole end of the inner barrel such that it also extends through the drill bit when the
inner and outer tubes are engaged. The barrel assembly is pressed against the hole bottom
which causes the clay to take an impression of the core stub left from the previous core
run. The inner barrel is then retrieved with the wire-line and a conventional barrel is
lowered to continue coring. At the completion of the run, the recovered core is fitted
together and the core is oriented by matching the piece of core from the upper end of the
core run with the oriented clay imprint. The clay barrel method can only be used in
inclined holes within the dip range of 45° to 70° where the weighted barrel will orient
itself as it is lowered down the hole. Like the Craelius method, this method works well
only if adjacent pieces of core can be matched.
Pendulum Orientation Method (Webber & Gowans, 1996) Orientation of the core is
based on orienting the last piece of core in each core run. The pendulum orientation
system incorporates a pendulum which moves under gravitational force while drilling to
indicate the lowest position at an inclined borehole (see Fig. 5.5). The system depends on
maintaining a fixed rotational relationship between the inner tube of the corebarrel and
the orientation device containing the pendulum. This is achieved by rigidly fixing the
orientation device to a modified spindle in the corebarrel head. Once the core run is
complete an overshot trigger is lowered to activate the core orientator. The overshot
device latches onto the core barrel assembly, triggering the pendulum by pushing it
downwards against the action of a spring. The lowest position of the inner tube in the
inclined borehole is then indicated by the point of the pendulum which emerges through
one of the 72 small holes on the indicator plate. The pendulum and the inner tube are then
fixed. The core barrel can then be removed. At the completion of the run, the last piece of
core can be oriented by marking the lowest point from the point of the pendulum and the
rest pieces of core can be oriented by matching the pieces of core from the lower end of
the core run. The system is designed to operate in boreholes with a minimum inclination
of 5° from the vertical. Like the Craelius method and clay barrel method, this method
works well only if adjacent pieces of core can be matched.
Fig. 5.3 A sketch showing stages in

removing an integral sample: (1)
positioning rod; (2) connecting sleeve;
(3) cementing material; and (4) integral
sample before drilling it free (after
Goodman, 1976).
Fig. 5.4 Clay core barrel used to orient

diamond core in an inclined drill hole
(after Call et al., 1981).
5.3.2 Inspecting borehole walls

The sidewalls of the borehole from which the core has been extracted offer a unique
picture of the subsurface where all structural features of the rock formation are still in
their original position. This view of the rock can be important when portions of rock core
have been lost during the drilling operation, particularly weak seam fillers, and when the
true dip and dip direction of the structural features are required.
The spacing of discontinuities can be determined by televiewing or photographing the
borehole walls. Experience in examining a variety of rock types drilled using different
methods suggests that, for most rock types, the drilling process does not create significant
fractures in the borehole wall (Gunning, 1992). As a result discontinuity spacing
collected from borehole walls may be inherently more accurate and precise than that
derived from borehole cores which can be badly affected by drilling and handling
process.
Fig. 5.5 A pendulum orientation device

attached to a conventional corebarrel
(after Webber & Gowans, 1996).
Borehole viewing and photography equipment include borescopes, photographic
cameras, TV cameras, sonic imagery loggers, caliper loggers, and alignment survey
devices.
Borehole Periscopes can be used in small holes, but due to distortion of the optical path
the depth is usually limited to about 30 m (Goodman, 1976; ISRM, 1978).
Borehole cameras can be used to take photos of the borehole wall and the orientation and
spacing data can be obtained by interpreting the photos (Goodman, 1976; ISRM, 1978).
Closed Circuit Television (CCTV) provides a means of directly inspecting a borehole wall
and if the direction of view of the camera can be orientated, it is possible to determine
discontinuity orientation and spacing. In addition aperture, infilling and water seepage
may also be assessed. Successful CCTV surveys can be conducted in either dry or water-
filled holes. For best results in either case the borehole wall should be clean and stable. If
the borehole is full of water then measures should be taken to ensure that the water is
clear enough to give a side view image of the borehole wall (Gunning, 1992; McMillan et
al., 1996).
Acoustic Televiewer and Dipmeter offer great potential for analyzing discontinuities.
Advances in technology and digital instrumentation mean that such methods can provide
effective data acquisition systems. Acoustic televiewers only operate in fluid filled
boreholes (Gunning, 1992).
5.3.3 Large-diameter borings

Large-diameter borings, 2 feet or more in diameter (e.g., the borings of large drilled
shafts), permit direct examination of the sidewalls and bottoms of the boring and provides
access for obtaining high-quality undisturbed samples. Direct inspection of the sidewalls
and bottoms may reveal details, such as thin weak layers or old shear planes, that may not
be detected by continuous undisturbed sampling. However, direct measurements may not
always be possible because of water in the borehole and concerns for the safety of
personnel.
It is very often that no outcrop can be used at a site to obtain trace lengths of
discontinuities. In this case, trace length data can be obtained from the sidewalls and
bottoms of large-diameter borings such as the borings of large drilled shafts. If only
several traces or even no trace is present at a bottom, one can use bottoms at different
depths during the drilling process to collect the trace length data and then use the entire
data set (see Fig. 5.6).
5.4 EXPLORATORY EXCAVATIONS
Test pits, test trenches, and exploratory tunnels provide access for larger-scaled
observations of rock mass conditions, for determining top of rock profile in highly
weathered rock/soil interfaces, and for some in situ tests which cannot be executed in a
smaller borehole.
5.4.1 Test pits and trenches

In weak or highly fractured rock, test pits and trenches can be constructed quickly and
economically by surface-type excavation equipment. Final excavation to grade where
samples are to be obtained or in situ tests performed must be done carefully. Test pits and
trenches are generally used only above the groundwater level. Exploratory trench
excavations are often used in fault evaluation studies. An extension of a bedrock fault
into much younger overburden materials exposed by trenching is usually considered
proof of recent fault activity.
Fig. 5.6 Sampling at the bottom of the

hole, during the drilling process at
different depths, to obtain trace length
data (L is the depth of the drilled
shaft).
5.4.2 Exploratory tunnels

Exploratory tunnels/adits permit detailed examination of the composition and geometry
of rock structures such as joints, fractures, faults, shear zones, and solution channels.
They are commonly used to explore conditions at the locations of large underground
excavations and the foundations and abutments of large dam projects. They are
particularly appropriate in defining the extent of marginal strength rock or adverse rock
structure suspected from surface mapping and boring information. For major projects
where high-intensity loads will be transmitted to foundations or abutments, tunnels/adits
afford the only practical means for testing in-place rock at locations and in directions
corresponding to the structure loading. The detailed geology of exploratory tunnels,
regardless of their purpose, should be mapped carefully. The cost of obtaining an
accurate and reliable geologic map of a tunnel is usually insignificant compared with the
cost of the tunnel. The geologic information gained from such mapping provides a very
useful additional dimension to interpretations of rock structure deduced from other
sources. A complete picture of the site geology can be achieved only when the geologic
data and interpretations from surface mapping, borings, and pilot tunnels are combined
and well correlated. When exploratory tunnels are strategically located, they can often be
incorporated into the permanent structure. Exploratory tunnels can be used for drainage
and post-construction observations to determine seepage quantities and to confirm certain
design assumptions. On some projects, exploratory tunnels may be used for permanent
access or for utility conduits.
5.5 GEOPHYSICAL EXPLORATIONS
5.5.1 General description

Geophysical techniques consist of making indirect measurements on the ground surface,
or in boreholes, to obtain generalized subsurface information. Geologic information is
obtained through analysis or interpretation of these measurements. Boreholes or other
subsurface explorations are needed for reference and control when geophysical methods
are used. Geophysical explorations are of greatest value when performed early in the field
exploration program in combination with limited subsurface explorations. Geophysical
explorations are appropriate for a rapid, though approximate, location and correlation of
geologic features such as stratigraphy, lithology, discontinuities, ground water, and for
the in situ measurement of dynamic elastic moduli and densities. The cost of geophysical
explorations is generally low compared with the cost of core borings or test pits, and
considerable savings may be realized by judicious use of these methods.
Geophysical methods can be classified as active or passive techniques. Active
techniques impart some energy or effect into the earth and measure the earth materials’
response. Passive measurements record the strengths of various natural fields which are
continuous in existence. Active techniques generally produce more accurate results or
more detailed solutions due to the ability to control the size and location of the active
source.
5.5.2 Seismic methods

Seismic methods are the most commonly conducted geophysical surveys for engineering
investigations. Seismic surveys measure the relative arrival times, and thus the velocity of
seismic waves traveling between an energy source and a number of geophones. The
energy source may be a hammer blow, an explosion of a propaneoxygen mixture in a
heavy chamber (gas-gun), or a light explosive charge. There are two major classes of
seismic waves: body waves which pass through the volume of a material and surface
waves that exist only near a boundary. The body waves consist of the compressional or
pressure or primary wave (P-wave) and the secondary or transverse or shear wave (S-
wave). P-waves travel through all media that support seismic waves. P-waves in fluids,
e.g. water and air, are commonly referred to as acoustic waves. S-waves travel slower
than P-waves and can only transit material that has shear strength. S-waves do not exist in
liquids and gases, as these media have no shear strength. The velocities of the P- and S-
waves are related to the elastic properties and density of a medium by the following
equations:
(5.4)
(5.5)
(5.6)
E=2G(1+ν)
(5.7)
where ν is the Poisson’s ratio; Vp is the velocity of the P-wave; Vs is the velocity of the S-
wave; G is the shear modulus; ρ is the density; and E is the Young’s or elastic modulus. It
should be noted that these are not independent equations. Knowing two velocities
uniquely determines only two unknowns of ρ, ν and E. Shear modulus is dependent on
two other values. Usually the possible range of ρ is approximated and ν is estimated. The
typical density values of intact rocks have been presented in Table 2.7. Table 5.1
provides some typical values of Vp and ν. The velocity of the S-wave in most rocks is
about half the velocity of the P-wave.
Surface waves are produced by surface impacts, explosions and wave form changes at
boundaries. One of the surface waves is the Rayleigh wave which travels about 10%
slower than the S-wave. The Rayleigh wave exhibits vertical and horizontal displacement
in the vertical plane of the ray path. A point in the path of a Rayleigh wave moves back,
down, forward, and up repetitively in an ellipse like ocean waves.
The equipment used for seismic surveys includes the following components:
1) Seismic sources. The seismic source may be a hammer repetitively striking an
aluminum plate or weighted plank, drop weights of varying sizes, a rifle shot, a
harmonic oscillator, waterborne mechanisms, or explosives. The energy disturbance
for seismic work is most often called the “shot,” an archaic term
Table 5.1 Typical/representative field values of Vp
and ν (after ASCE, 1998).
Material Vp (m/s) ν
Air 330
Damp loam 300–750
Dry sand 450–900 0.3–0.35
Clay 900–1,800 ~0.5
Fresh, shallow water 1,430–1,490
Saturated, loose sand 1,500

Basal/lodgement till 1,700–2,300
Rock 0.15–0.25
Weathered igneous & metamorphic rock 450–3,700
Weathered sedimentary rock 600–3,000
Shale 800–3,700
Sandstone 2,200–4,000
Metamorphic rock 2,400–6,600
Unweathered bsalt 2,600–4,300
Dolostone and limestone 4,300–6,700
Unweathered granite 4,800–6,700
Steel 6,000
from petroleum seismic exploration. Reference to the “shot” does not necessarily
mean an explosive or rifle source was used. The type of survey dictates some
source parameters. Smaller mass, higher frequency sources are preferable. Higher
frequencies give shorter wavelengths and more precision in choosing arrivals and
estimating depths. Yet sufficient energy needs to be entered to obtain a strong
return at the end of the survey line.
2) Geophones. The geophones receiving seismic energy are either accelerometers or
velocity transducers, and convert ground shaking into a voltage response. Most
geophones are vertical, single-axis sensors to receive the incoming wave form from
beneath the surface. Some geophones have horizontal-axis response for S-wave or
surface wave assessments. Triaxial phones, capable of measuring absolute response,
are used in specialized surveys. Geophones are chosen for their frequency band
response.
3) Seismographs. The equipment that records input geophone voltages in a timed
sequence is the seismograph. Current practice uses seismographs that store the
channels’ signals as digital data in discrete time units. Earlier seismographs would
record directly to paper or photographic film. Stacking, inputting, and processing the
vast volumes of data and archiving the information for the client virtually require
digital seismographs.
In a homogeneous medium a bundle of seismic energy travels in a straight line. Upon
striking a boundary between different material properties, wave energy is refracted,
reflected, and converted. The properties of the two media and the angle at which the
incident ray path strikes will determine the amount of energy reflected off the surface,
refracted into the adjoining material, lost as heat, and changed to other wave types.
Figure 5.7 shows the refraction and reflection of a seismic ray incident at an angle θ1 on
the boundary between media with velocities V1 and V2. Refraction, as with any wave,
obeys Snell’s Law relating the angle between the ray path and the normal to the boundary
to the velocity V(VP or Vs appropriate). Thus in Figure 5.7(a), we have
(5.8)
If refraction continues through a series of such interfaces parallel to each other, we have
(5.9)
If the lower material has a higher velocity (V2>V1 in Fig. 5.7), a particular down-going
ray making an angle
(5.10)
with the normal will critically refract along the boundary and return to the surface at the
same angle [see Fig. 5.7(b)].
Fig. 5.7 (a) Reflection and refraction of

a seismic ray incident at an angle θ1 on
the boundary between media with
velocities V1 and V2; and (b) Critically
refracted wave travels below the
boundary and returns to the surface.
(a) Seismic refraction method

The method of seismic refraction is schematically shown in Figure 5.8. Waves of
different types traveling by various paths to points on the surface at various horizontal
distances, X, from the shot are detected by geophones. For geophones near the shot, the
first arrivals will be directly from the shot. If the lower material has a higher velocity
[V2>V1 in Fig. 5.8(a)], rays traveling along the boundary will be the first to arrive at
geopehone away from the shot. If the time of first arrivals is plotted against distance X, a
plot with two straight branches as shown in Figure 5.8(b) will be obtained. From an
examination of Figure 5.8(b) one can obtain the following information:
1) The slopes of the two straight lines are equal to 1/V1 and 1/V2, respectively.
2) The depth to the interface, D, can be obtained by

seismic refraction method: (a) Shot,
geophones and direct and refiracted
wave paths; and (b) Time versus
distance plot for seismic refraction
survey as shown in (a).
(5.11)
where XC is the crossover distance; and D is the depth to the horizontal refracting
interface (Fig. 5.8).
When the interface is dipping downwards from the shot towards the geophones, the
velocity of the lower medium obtained as described above will be smaller than the true
velocity. In the opposite situation, with the interface rising from the shot towards the
geophones, the obtained velocity will be higher than the true one. By reversing shots and
measuring the velocities in both directions (up- and down- dip) the dip of the interface
can be estimated (ASCE, 1998).
The method described above for finding the seismic wave velocities and the depths to
the refracting interfaces can readily be extended to systems with three or more layers with
boundaries that need not be planar and velocities that may show lateral changes. For
details, the reader can refer to Griffiths and King (1981) and ASCE (1998).
In simple cases, such as the two layer system described above, the seismic refraction
method can predict depths to geological surfaces with an accuracy of ±10%. In complex
formation, the accuracy drops considerably, and is much more dependent on the skill of
the operators. The two most difficult geologic conditions for accurate refraction work are
the existence of a thin water-saturated zone just above the bedrock and the existence of a
weathered zone at the top of bedrock. The method fails completely, however, when a
high velocity layer covers a low velocity one, since there is no refraction at this case.
(b) Seismic reflection method

A portion of the seismic energy striking an interface between two differing materials will
be reflected from the interface (Fig. 5.7). The ratio of the reflected energy to incident
energy is called the reflection coefficient. The reflection coefficient is defined in terms of
the densities and seismic velocities of the two materials as:
(5.12)
where R is the reflection coefficient; ρ1 and ρ2 are densities respectively of the first and
second layers; V1 and V2 are seismic velocities respectively of the first and second layers.
Modern reflection methods can ordinarily detect isolated interfaces whose reflection
coefficients are as small as 0.02.
The physical process of reflection is illustrated in Figure 5.9, where the ray paths from
the successive layers are shown. As in Figure 5.9, there are commonly several layers
beneath the ground surface which contribute reflections to a single seismogram. Thus,
seismic reflection data are more complex than refraction data because it is these later
arrivals that yield information about the deeper layers. At later times in the record, more
noise is present thus making the reflections difficult to extract from the unprocessed
record. Figure 5.10 indicates the paths of arrivals that would be recorded on a multi-
channel seismograph. Another important feature of modern reflection data acquisition is
illustrated by Figure 5.11. If multiple shots, S1 and S2, are recorded by multiple
geophones, G1 and G2, and the geometry is as shown in the figure, the reflector point for
both rays is the same. However, the ray paths are not the same length, thus the reflection
will occur at different times on the two traces. This time delay, whose magnitude is
indicative of the subsurface velocities, is called normalmoveout. With an appropriate
time shift, called the normal-moveout correction, the two traces (S1 to G2 and S2 to G1)
can be summed, greatly enhancing the reflected energy and canceling spurious noise.
Fig. 5.9 Schematic of seismic

reflection method.
Fig. 5.10 Multi-channel recording of

seismic reflections.
Fig. 5.11 Common depth point

recording.
(c) Rayleigh wave method

Of the surface waves, the Rayleigh wave is important in engineering studies because of
its simplicity and because of the close relationship of its velocity to the shear wave
velocity for earth materials. Approximation of Rayleigh wave velocities as shear-wave
velocities causes less than a 10-percent error.
Rayleigh wave studies for engineering purposes have most often been made in the past
by direct observation of the Rayleigh wave velocities. One method consists of excitation
of a monochromatic wave train and the direct observation of the travel time of this wave
train between two points. As the frequency is known, the wavelength is determined by
dividing the velocity by the frequency.
(d) Cross hole method

Cross hole testing takes advantage of generating and recording seismic body waves, both
the P- and S-waves, at selected depth intervals where the source and receiver(s) are
maintained at equal elevations for each measurement. Figure 5.12 illustrates a general
field setup for the cross hole seismic test method. Knowing the distance between the
source borehole and the geophone borehole and measuring the time of travel of the
induced wave from the source to the geophone, the velocity of the rock mass can be
simply obtained by
(5.13)
where D (D1 or D2 in Fig. 5.12) is the distance between the source borehole and the
geophone borehole; and t is the time of travel of the induced wave from the source to the
geophone.
Particle motions generated with different seismic source types used during cross hole
testing are three-directional. Therefore, three-component geophones with orthogonal
orientations yield optimal results when acquiring cross hole P- and/or S-wave seismic
signals. The requirement for multiple drill holes in cross hole testing means that care
must be taken when completing each borehole with casing and grout. ASTM procedures
call for PVC casing and a grout mix that closely matches the formation density. Another
critical element of cross hole testing, which is often ignored,
Fig. 5.12 Schematic of cross hole

method.
is the requirement for borehole directional surveys. Borehole verticality and direction
(azimuth) measurements should be performed at every depth interval that seismic data are
acquired. Since seismic wave travel times should be measured to the nearest tenth of a
millisecond, relative borehole positions should be known to within a tenth of a foot.
Assuming that the boreholes are vertical and plumb leads to computational inaccuracies
and ultimately to data which cannot be quality assured.
Unlike surface seismic techniques previously described, cross hole testing requires a
more careful interpretation of the wave forms acquired at each depth. For example, in
cross hole testing, the first arrival is not always the direct ray path. As illustrated
schematically in Figure 5.13, when the source and receivers are located within a layer
that has a lower velocity than either the layer above or below it (this is termed a hidden
layer in refraction testing), refracted waves can be the first arrivals. Both the
source/receiver distance above or below the high velocity layer and the velocity contrast
across the seismic interface determine if the refracted wave will arrive before the direct
wave.
(e) Down hole method and up hole method

For the down hole method, the seismic waves are generated by a source located at the
ground surface and the geophones are located inside a borehole drilled adjacent to the
energy source (see Fig. 5.14). Each test involves the determination of the wave velocity
based on the distance between the ground surface and the level at which the geophone is
located, and the respective traveling time.
The equipment is basically the same as for the cross hole method. The only significant
difference is the energy source. The impulses are generated by hammering a plate after
assuring a good contact with the ground. For generating P-waves the blow is normal to
the surface of the plate (and to the ground surface) and for generating a S-waves the blow
is horizontal (parallel to the ground surface). When measuring S-wave velocities, the test
can be repeated with the geophone at the same level, by reversing the direction of the
impact, which allows a second recording.
In the opposite, for the up hole method, the seismic waves are generated in the
borehole and the geophones are located at the ground surface.
The main advantage of the down hole and up hole methods over the cross hole method
lies in the fact that only one regular borehole is required to perform the test.
Fig. 5.13 Refracted ray paths in cross

hole seismic test where V1>V2<V3 and
V1<V3.
Fig. 5.14 Schematic of down hole

method.
5.5.3 Electrical resistivity method

Electrical resistivity surveying is based on the principle that the distribution of electrical
potential in the ground around a current-carrying electrode depends on the electrical
resistivities and distribution of the surrounding soils and rocks. The usual practice in the
field is to apply an electrical direct current (DC) between two electrodes implanted in the
ground and to measure the difference of potential between two additional electrodes that
do not carry current (see Fig. 5.15). Usually, the potential electrodes are in line between
the current electrodes, but in principle, they can be located anywhere. The current used is
direct current, commutated direct current (i.e., a square-wave alternating current), or AC
of low frequency (typically about 20 Hz). All analysis and interpretation are done on the
basis of direct currents. The distribution of potential can be related theoretically to ground
resistivities and their distribution for some simple cases, notably, the case of a
horizontally stratified ground and the case of homogeneous masses separated by vertical
planes (e.g., a vertical fault with a large throw or a vertical dike). For other kinds of
resistivity distributions, interpretation is usually done by qualitative comparison of
observed response with that of idealized hypothetical models or on the basis of empirical
methods.
Mineral grains composing soils and rocks are essentially nonconductive, except in
some exotic materials such as metallic ores, so the resistivity of soils and rocks is
governed primarily by the amount of pore water, its resistivity, and the arrangement of
the pores. Since the resistivity of a soil or rock is controlled primarily by the pore water
conditions, there are wide ranges in resistivity for any particular soil or rock type (see
Table 5.2), and resistivity values cannot be directly interpreted in terms of soil type or
lithology. Commonly, however, zones of distinctive resistivity can be associated with
specific soil or rock units on the basis of local field or drill hole information, and
resistivity surveys can be used profitably to extend field investigations into areas with
very limited or nonexistent data. Also, resistivity surveys may be used as a
reconnaissance method, to detect anomalies that can be further investigated by
complementary geophysical methods and/or drill holes.
Fig. 5.15 Schematic of electrical

resistivity method.
Table 5.2 Typical resistivity values of soils and
rocks (after Griffiths & King, 1981).
Soil/Rock Resistivity (ohm-m)
Soft shale, clay 1–10
Hard shale 10–500
Sand 50–1,000
Sandstone 50–5,000
Porous limestone 100–5,000
Dense limestone >5,000
Igneous rock 100–1,000,000
Metamorphic rock 50–1,000,000
5.5.4 Ground penetration radar

Ground Penetrating Radar (GPR), also known as ground probing radar, ground radar or
georadar, has been widely used in high-resolution mapping of soil and rock stratigraphy
(Deng, 1996; Sharma, 1997). The GPR method uses high-frequency (80 to 1,000 MHz)
electromagnetic (EM) waves transmitted from a radar antenna to probe the earth. The
transmitted EM waves are reflected from various interfaces within the ground and are
detected by the radar receiver. Reflecting interfaces may be soil horizons, the
groundwater surface, soil/rock interfaces, man-made objects, or any other interfaces
possessing a contrast in dielectric properties. The GPR method is analogous to seismic
reflection except for the energy source (Sharma, 1997; ASCE, 1998).
Contrasts in dielectric properties across an interface cause EM waves to be reflected.
Fracture fillings with dielectric properties different from their adjacent rock materials can
cause radar reflections and thus can be detected.
One limitation of the GPR method is that the penetration depth of radar is limited
usually less than 20 meters (Cummings, 1990; Kearey & Brooks, 1991; Sharma, 1997;
ASCE, 1998). At the Gypsy Outcrop Site in Northeastern Oklahoma, the maximum depth
with noticeable radar response is about 10 meters (Deng, 1996). Therefore, the GPR
method can only be used for shallow depth survey.
Similar to the seismic wave method, the processing and interpretation of recorded
GPR data is critically important. Due to the kinematic similarities between radar and
seismic wave propagation, seismic processing techniques are widely used to process the
GPR data (Deng, 1996; Sharma, 1997).
5.6 LABORATORY TESTING
Laboratory tests are usually performed to determine index values for identification and
correlation, further refining the geologic model of the site, and provide values for
engineering properties of the rock used in the analysis and design of foundations. The
selection of samples and the number and type of tests are influenced by local subsurface
conditions and the size and type of structure. Prior to any laboratory testing, rock cores
should have been visually classified and logged.
Selection of samples and the type and number of tests can best be accomplished after
development of the geologic model using results of field observations and examination of
rock cores, together with other geotechnical data obtained from earlier preliminary
investigations. The geologic model, in the form of profiles and sections, will change as
the level of testing and the number of tests progresses. Testing requirements are also
likely to change as more data become available and are reviewed for project needs.
Table 5.3 summarizes laboratory tests according to purpose and type. The tests listed
are the types more commonly performed for input to rock foundation analysis and design.
Details and procedures for individual test types can be found in books on rock mechanics
and rock engineering.
For rock specimens with the same geometrical shape, the strength decreases with
increasing size, reaching a limit value asymptotically (see Section 4.5 for details on scale
effect). This size, beyond which no further decrease in strength is observed, depends on
the type of rock material. A simplified explanation for this phenomenon is that rock is not
a continuous solid material, but may contain various types of discontinuities or flaws.
The strength of any rock specimen is, therefore, a statistical value depending on how
many and what type of discontinuities are present. In smaller specimens the probability of
the presence of such discontinuities is smaller and thus the strength is higher.
In addition to the size effect, the strength of a rock specimen is affected by its shape,
i.e. the length-to-diameter ratio of the test specimen or the width-to-height ratio of the
specimen with a square cross-section. Figure 5.16 shows the effect of length-to-
Table 5.3 Summary of purpose and type of
laboratory tests for rock (after ASCE, 1996).
Purpose of test Type of test
Strength Unconfined compression
Direct shear
Triaxial compression
Direct tension
Brazilian split
Point load1
Deformability Unconfined compression
Triaxial compression
Swell
Creep
Permeability Gas permeability
Characterization Water content
Porosity
Density (unit weight)
Specific gravity
Absorption
Rebound
Sonic velocities
Abrasion resistance
1. Point load tests are also frequently performed in the field.
diameter ratio on the unconfined compressive strength of cylindrical sandstone specimens

(John, 1972). This effect may be explained by the variation in the stress distribution in
the test specimen as a result of the end constraint. The influence of loading platens on the
specimen ends in a compression test diminishes with increasing length of specimen.
To minimize the effects of size and shape on the test results, minimum dimensions and
minimum height-to-diameter ratios for test specimens have been recommended by ASTM
(1971) and ISRM (1979a).
5.7 IN SITU TESTING
5.7.1 General description

In situ tests are often the best means for determining the engineering properties of
subsurface materials and, in some cases, may be the only way to obtain meaningful
results. Table 5.4 lists some of the most widely used in situ tests and their purposes. In
situ rock tests are performed to determine in situ stresses and deformation properties of
the jointed rock mass, shear strength of jointed rock mass or critically weak seams within
the rock mass, residual stresses within the rock mass, and rock mass permeability. Large-
scale in situ tests tend to average out the effect of complex interactions. In situ tests in
rock are frequently expensive and should be reserved for projects with large,
concentrated loads. Well-conducted tests may be useful in reducing overly conservative
assumptions. Such tests should be located in the same general area as the proposed
structure and test loading should be applied in the same direction as the proposed
structural loading.
Some of the strength and deformability tests are described in more detail in the
following.
Fig. 5.16 Influence of length-to-

diameter ratio on the unconfined
compressive strength of dry sandstone
(after John, 1972).
Table 5.4 Summary of purpose and type of in situ

tests for rock (modified from ASCE, 1996).
Purpose of test Type of test
Strength Field vane shear1
Direct shear
Pressuremeter2
Unconfined compression2
Borehole jacking2
Deformability Seismic3
Pressuremeter or dilatometer
Plate bearing
Radial (tunnel) jacking2
Borehole jacking2
Chamber (gallery) pressure2
Bearing capacity Plate bearing1
Standard penetration1
Stress conditions Hydraulic fracturing
Pressuremeter
Overcoring
Flat jack
Uniaxial (tunnel) jacking2
Chamber (gallery) pressure2
Permeability Constant head
Rising or falling head
Well slug pumping
Pressure injection
Notes: 1. Primarily for clay shales, badly decomposed, or moderately soft rocks, and rock with soft
seams.
2. Less frequently used.
3. Dynamic deformability.
5.7.2 Strength tests

The most common in situ test for determining the strength of rock masses is the direct
shear test. Triaxial tests have been conducted in particular situations, but due to
difficulties in execution, they are of very restricted use. The scope of the direct shear test
is to measure the peak and residual shear strength as a function of the normal stress on
the sheared plane. At least three or four specimens should be tested at different normal
stresses on the same test horizon.
Figure 5.17 shows a typical in situ direct shear test arrangement. In the case of tests
conducted in adits, the reaction for the normal load is obtained from the opposite wall of
the adit. Tests can be conducted on a rock surface using cables anchored into the rock
adjacent to the test site to supply the normal reaction. The procedure for conducting an in
situ direct shear test can be referred to Oliveira and Charrua Graca (1987). By conducting
direct shear test on a set (three or four) of blocks, the shear strength curves as shown in
Figure 5.18 can be obtained. The peak and residual strength
Fig. 5.17 Schematic of in situ shear

test (after Oliveira & Charrua Graca,
1987).
Fig. 5.18 Shear strength curves from in

situ shear tests (after Oliveira &
Charrua Graca, 1987).
parameters (cohesion and friction angle) can then be determined from the shear strength
curves.
5.7.3 Deformability tests

As shown in Table 5.4, there are a number of in situ test methods available to estimate
rock modulus (ASCE, 1996). The details about the seismic methods have been discussed
in Section 5.5. In this section, the borehole dilatometer test, the boehole jack test, the
plate bearing test, the flat jack test and the radial jacking test will be discussed.
(a) Borehole dilatometer test

The borehole dilatometer test expands a fluid filled flexible membrane in a borehole
causing the surrounding wall of rock to deform. The fluid pressure and the volume of
fluid equivalent to the volume of displaced rock are recorded. From the theory of
elasticity, pressure and volume changes are related to the modulus. The primary
advantage of the borehole dilatometer test is its low cost. The test is, however, restricted
to relatively soft rock. Furthermore, the test influences only a relatively small volume of
rock. Hence, modulus values derived from the borehole dilatometer tests are not
considered to be representative of rock mass conditions.
The deformation modulus Em of the rock mass can be calculated from the dilatometer
test by (Goodman, 1980)
(5.14)
where νm is the Poisson’s ratio of the rock mass; d is the diameter of the borehole test
section; ∆p is the change in pressure applied uniformly over the borehole surface; and ∆d
is the measured radial deformation.
For a Colorado School of Mines (CSM) dilatometer, a calibration test in a material of
known modulus need be conducted to determine the stiffness of the membrane system.
Figure 5.19 shows typical pressure-dilation curves for a calibration test and a test carried
out in rock. A complete test usually consists of three loading and unloading cycles, with
dilation and pressure readings being taken on both the loading and unloading cycles.
The shear modulus Gm and the deformation modulus Em of the rock mass in the
borehole test section are given by (ISRM, 1987)
(5.15)
and
Em=2(1+νm)Gm
(5.16)
where L is the length of the test section (cell membrane); d is the diameter of the borehole
test section; νm is the Poisson’s ratio of the rock mass; ρ is the pump constant (the fluid
volume displaced per turn of pump wheel); and km is the stiffness of the rock mass, which
can be obtained by
Fig. 5.19 Typical pressure-dilation

curves for a Colorado School of Mines
(CSM) dilatometer (IRSM, 1987).
(5.17)
where ks is the stiffness of the hydraulic system [Equation (5.19)]; and kT is the stiffness
of the overall system plus the rock mass (ratio D/C in Fig. 5.19). The rock mass stiffness
km is calculated from calibration of the hydraulic system and the results of a pressure-
dilation test carried out in a calibration cylinder of known modulus. The steps for
calculating the rock mass stiffness are as follows.
If the shear modulus and Poisson’s ratio of the calibration cylinder are respectively Gc
and νc, the stiffness of the calibration cylinder kc is
(5.18)
where ri and ro are respectively the inside and outside radii of the calibration cylinder.
The stiffness of the hydraulic system ks is calculated from the stiffness of the
calibration cylinder and the slope of the calibration pressure-dilation curve km (ratio B/A
in Fig. 5.19) as follows
(5.19)
It is also necessary to make a correction for pressure losses due to the rigidity of the
membrane. This is determined by inflating the dilatometer in the air without confinement
to show the pressure required to inflate the membrane and the hydraulic system.
pi,corr=pi−nmp (MPa)
(5.20)
where pi,corr is the corrected pressure; pi is the indicated pressure; n is the number of turns
to attain pi; and mp is the slope of pressure-dilation curve for dilation in air (MPa/turn).
Another correction is required to account for loss of volume in the hydraulic system
that takes place in inflating and seating the membrane. For the test measurements shown
in Figure 5.19, the net corrected number of turns ∆ncorr is calculated from
(5.21)
(b) Borehole jack test

Instead of applying a uniform pressure to the full cross-section of a borehole as in the
borehole dilatometer tests, the borehole jack presses plates against the borehole walls
using hydraulic pistons, wedges, or flat jacks. This technique allows the application of
significantly higher pressures required to deform hard rock. The NX-borehole (76 mm in
diameter) jack (also known as the hard-rock jack or Goodman Jack) is the best known
device for this test (Fig. 5.20). The deformation modulus of the rock mass can be
calculated from a NX-borehole jack test by (Heuze, 1984; Heuze & Amadei, 1985)
Ecalc=0.86(0.93)d(∆Qh/∆d)T*
(5.22)
where 0.86 is the factor for the three-dimensional effect; 0.93 is the hydraulic efficiency;
d is the diameter of the borehole; ∆d is the change of borehole diameter; ∆Qh is the
increment of hydraulic-line pressure; and T* is a coefficient depending on the Poisson’s
ratio νm of the rock mass (Table 5.5).
In rock with a deformation modulus greater than about 7 GPa, there will be a
longitudinal outward bending of the jack platens and the calculated modulus need be
corrected to obtain the true modulus Em (Fig. 5.21). This correction is necessary because
the bending gives larger displacements at the ends than at the center of the loading
platens and the displacement gauges are located near the ends of the platens (Heuze,
1984).
The advantage of the borehole jack test over the borehole dilatometer test is that the
unidirectional pressure can be imposed in a given orientation. The limitation of the
borehole jack test is that only a point modulus (for a small volume of rock mass) can be
obtained.
Fig. 5.20 Schematic of loading of NX-

borehole jack (Heuze, 1984).
Table 5.5 T* for full platen and rock contact (after
Heuze, 1984).
νm 0.1 0.2 0.25 0.3 0.33 0.4 0.5
T* 1.519 1.474 1.438 1.397 1.366 1.289 1.151
Fig. 5.21 Curve of Ecalc versus Em

(after Heuze, 1984).
(c) Plate bearing test

In a plate bearing test a plane circular area of the rock surface is loaded and the induced
deformation of the rock is measured. Usually the test is performed in small tunnels or
adits and two opposite surfaces are loaded, thus providing the reaction support for the
forces employed (Fig. 5.22). When it is necessary to conduct a test at the ground surface,
special structures such as cables anchored at some depth below the surface must be
employed to support the reaction (Fig. 5.23).
The site selected for a test should be large enough and carefully prepared. The areas to
be loaded and their vicinities, from 0.5 to 1 diameters of the loading plate, must be
cleaned of all disturbed rock. These areas should then be hand-prepared to become plane
and parallel.
Usually a tunnel diameter gauge connecting the centers of the loaded surface measures
the relative deformation. If more sophisticated information on deformation is needed,
multiple-position borehole extensonmeters (MPBX) installed in holes drilled into the
rock along the load axis can be used. The depth of the extensometer holes must be such
that the deepest anchor is beyond the zone of deformation, a distance of about six
diameters of the loading plate [Fig. 5.22(b)].
Assuming that the loaded surface behaves like a homogeneous infinite half space and
that the rock mass behaves like an isotropic elastic linear medium, the deformation
modulus of the rock mass can be calculated from the deformation measurements. For a
test condition in which the bearing plate is circular and has a circular hole in the center
through which the deformation measurements are made, the deformation modulus Em at
any depth z is given by the following expression
(5.23)
where δz is the measured displacement at depth z below the lower surface of the loading
plate; p is the applied pressure on the loading plate; νm is the Poisson’s ratio of the rock
mass; R is the outer radius of the loading plate; r is the radius of the hole in the center of
the loading plate; and C is a constant. For a perfectly rigid loading plate, the theoretical
solution gives C as π/2. Since the actual loading plate has some flexibility, the measured
deformation is somewhat greater than the theoretical deformation. This results in the
calculated deformation modulus being smaller than the true modulus and for this reason
the constant C is usually given the value of 2.
For a loading plate with no center hole, the deformation modulus is given by
(5.24)
For measurements at the surface of the rock where z=0, this expression reduces to
Fig. 5.22 (a) Simple plate bearing test

in a gallery; and (b) Plate bearing test
in a gallery using multiple-position
borehole extensonmeters (MPBX)
(after IRSM, 1979b).
Fig. 5.23 (a) Plate bearing test using a

single anchorage cable; and (b) Plate
bearing test using multiple anchorage
cables.
(5.25)
(d) Flat jack test

The flat jack test is a simple test in which slots cut in the rock mass are uniformly loaded
by flat jacks inserted into them (Fig. 5.24). Deformation of the rock mass caused by
pressurizing the flat jack is measured by the volumetric change in the jack fluid. The
deformation modulus of the rock mass is derived from relationships between jack
pressure and deformation.
Using loading, unloading and reloading cycles permits calculation of the deformation
modulus of the rock mass by (Jaeger & Cook, 1979)
(5.26)
where p is the applied pressure; 2c is the length of the jack; 2∆y is the variation of pin
separation; νm is the Poisson’s ratio of the rock mass; and y is the distance from the jack
center to each of a pair of measuring pins.
The primary advantages of the flat jack test lie in its ability to load a large volume of
rock and its relatively low cost.
Fig. 5.24 Flat jack test: (a) Front view;

and (b) Section after flat jack
installation (after Jaeger & Cook,
1979).
(e) Radial jacking test

In this test the load is applied uniformly to the complete surface of a test chamber with a
circular cross section and the radial deformations are measured along a number of axes.
The results of this test provide the deformation modulus of a larger volume of the rock
mass than is possible with the tests described earlier and, in the case of anisotropic rock,
can show the variation of the modulus with orientation. Although more information on
rock conditions is provided by the radial jacking test than the tests described earlier, the
high cost and time required to conduct the test means that very few will be carried out
and thus the results may not be representative of the overall site.
The best known and oldest variant of the radial jacking test is the pressure chamber
test in which the pressure is applied by means of water filling the chamber. With this
method, special care must be taken not to allow the water to flow from the chamber.
6
Axial load capacity of drilled shafts in rock
6.1 INTRODUCTION
The design of axially loaded drilled shafts in rock usually involves computation of
ultimate load capacity and prediction of settlement under working load. This chapter
addresses the determination of the ultimate load capacity while the prediction of
settlement at the working load will be discussed in Chapter 7.
Axially loaded drilled shafts in rock are designed to transfer structural loads to rock in
one of the following three ways (CGS, 1985):
1. Through side shear only;
2. Through end bearing only;
3. Through the combination of side shear and end bearing.
Situations where support is provided solely by side shear resistance are those where the
base of the drilled hole cannot be cleaned so that it is uncertain if any end bearing
resistance will be developed. Alternatively, where sound bedrock underlies low strength
overburden material, it may be possible to achieve the required support in end bearing
only, and assume that no side shear support is developed in the overburden. However,
where the shaft is drilled some depth into sound rock, a combination of side shear
resistance and end bearing resistance can be assumed (Kulhawy & Goodman, 1980).
The load bearing capacity of a drilled shaft in rock is determined by the smaller of the
two values: the structural strength of the shaft itself, and the ability of the rock to support
the loads transferred by the shaft.
6.2 CAPACITY OF DRILLED SHAFTS RELATED TO

REENFORCED CONCRETE
Axially loaded drilled shafts may fail in compression or by buckling. Buckling is possible
in the long and slender part that extends above the ground surface. Scour of the soil/rock
around the shaft will expose portions of the shaft, thus extending the unbraced length and
making the shaft more prone to buckling.
The capacity of a shaft as a reinforced concrete element is a function of the shaft
diameter, the strength of the concrete and the amount and type of reinforcement. The
shaft should be designed such that the working stresses are limited to the allowable
concrete stresses as shown in Table 6.1. For the reinforcing steel, the allowable design
stress should not exceed 40% of its specified minimum yield strength, nor 206.8 MPa
(30,000 psi) (ASCE, 1997).
Axial load capacity of drilled shafts in rock 215
In LRFD, the ultimate (factored) axial capacity of a drilled shaft can be calculated
using the expression for reinforced concrete columns:
(6.1)
where is the capacity reduction (resistance) factor=0.75 for spiral columns and 0.70
for horizontally tied columns (ACI, 1995); Qu is the nominal (computed) structural
capacity; β is the eccentricity factor=0.85 for spiral columns and 0.80 for tied columns;
is the specified minimum concrete strength; Ac is the cross-sectional area of the
concrete; fy is the yield strength of the longitudinal reinforcing steel; and As is the cross-
sectional area of the longitudinal reinforcing steel.
The Standard Specifications for Highway Bridges adopted by the American
Association of State Highway and Transportation Officials (AASHTO, 1989) stipulates a
minimum shaft diameter of 18 inches, with shaft sizing in 6-inch increments. Where the
potential for lateral loading is not significant, drilled shafts need to be reinforced for axial
loads only. The design of longitudinal and spiral reinforcement should conform to the
requirements of reinforced compression members.
Table 6.1 Allowable concrete stresses for drilled
shafts (after ASCE, 1993).
Uniform axial compression
Confined 0.33f′c
Unconfined 0.27f′c
Uniform axial tension 0
Bending (extreme fiber)
Compression 0.40f′c
Tension 0
Note: f′c is the specified minimum concrete strength.
6.3 CAPACITY OF DRILLED SHAFTS RELATED TO ROCK
Assuming that the shaft itself is strong enough, its load capacity depends on the capacity
of the rock to accept without distress the loads transmitted from the shaft. The required
area of shaft-rock interface (i.e., the size of drilled shaft) depends on this factor. The
ultimate axial load of a drilled shaft related to rock, Qu, consists of the ultimate side shear
load, Qus, and the ultimate end bearing load, Qub (see Fig. 6.1):
Qu=Qus+Qub
(6.2)
The ultimate side shear load and the ultimate end bearing load are respectively calculated
as the average side shear resistance multiplied by the shaft side surface area and as the
end bearing resistance multiplied by the shaft bottom area, i.e.
Fig. 6.1 Axially loaded drilled shaft.

Qus=πBLτmax
(6.3)
(6.4)
where L and B are respectively the length and diameter of the shaft; and τmax and qmax are
respectively the average side shear resistance and the end bearing resistance.
The ultimate side shear resistance and the end bearing resistance are usually
determined based on local experience and building codes, empirical relations, or field
load tests. Methods based on local experience and building codes and empirical relations
are discussed in this chapter. The methods for conducting field load tests and
interpretation of test results will be discussed in Chapter 12.
6.3.1 Side shear resistance

The shear resistance mobilized at the shaft-rock interface is affected by many factors.
These include the shaft roughness, strength and deformation properties of the concrete
and the rock mass, geometry of the shaft, and initial stresses in the ground. The effect of
shaft roughness is emphasized by most investigators and considered in a number of
empirical relations for estimating the side shear resistance.
(a) Correlation with SPT N value

Standard Penetration Tests (SPT) are often carried out in weak or weathered rock. Table
6.2 shows the measured side shear resistances of drilled shafts and their corresponding
SPT N values in weathered sedimentary rocks. It can be seen that the τmax/N ratio is
generally smaller than 2.0 except the case reported by Toh et al. (1989). We can also see
that the τmax/N ratio tends to decrease as N increases.
Table 6.2 Side shear resistance and SPT N values in
weathered sedimentary rock.
Rock SPT N values τmax τmax/N Reference
(blows/0.3 m) (kPa) (kPa)
Highly weathered siltstone 230 >195– >0.87– Buttling (1986)
226 1.0
Highly weathered siltstone, silty 100–180 100– 1.0–1.8 Chang and Wong
sandstone and shale 320 (1987)
Very dense clayey/sandy silt to 110–127 80–125 0.63– Buttling and Lam
highly weathered siltstone 1.14 (1988)
Highly to moderately weathered 200–375 340 0.9–1.7
siltstone
Completely to partly weathered 100–150 – 1.2–3.7 Toh et al. (1989)
interbedded sandstone, siltstone and 150–200 – 0.6–2.3
shale/mudstone
Highly to moderately fragmented 400–1000 300– 0.5–0.8 Radhakrishnan and
siltstone/shale 800 Leung (1989)
Highly weathered sandy shale 150–200 120– 0.8–0.7 Moh et al. (1993)
140
Slightly weathered sandy shale and 375–430 240– ave. 0.65
sandstone 280
(b) Empirical relations between side shear resistance and unconfined

compressive strength of intact rock
Empirical relations between the side shear resistance and the unconfined compressive
strength of rock have been proposed by many researchers. The form of these empirical
relations can be generalized as
τmax= ασcβ
(6.5)
where τmax is the side shear resistance; σc is the unconfined compressive strength of the
intact rock (if the intact rock is stronger than the shaft concrete, σc of the concrete is
used); and α and β are empirical factors.
The empirical factors proposed by a number of researchers have been summarized by
O’Neill et al. (1996) and are shown in Table 6.3. Most of these empirical relations were
developed for specific and limited data sets, which may have correlated well with the
proposed equations. However, O’Neill et al. (1996) compared the first nine empirical
relations listed in Table 6.3 with an international database of 137 pile load tests in
intermediate-strength rock and concluded that none of the methods could be considered a
satisfactory predictor for the database.
Kulhawy and Phoon (1993) developed a relatively extensive load test database for
drilled shafts in soil and rock and presented their data both for individual shaft load tests
and as site-averaged data. The results are shown in Figures 6.2 and 6.3, in terms of
adhesion factor, σc, versus normalized shear strength, cu/pa or σc/2pa (assuming cu≈ σc/2),
where pa is atmospheric pressure (≈0.1 MPa). It should be noted that Kulhawy and
Table 6.3 Empirical factors a and β for side shear
resistance (modified from O’Neill et al., 1996).
Design method α β
Horvath and Kenney (1979) 0.21 0.50
Carter and Kulhawy (1988) 0.20 0.50
Williams et al. (1980) 0.44 0.36
Rowe and Armitage (1984) 0.40 0.57
Rosenberg and Journeaux (1976) 0.34 0.51
Reynolds and Kaderbek (1980) 0.30 1.00
Gupton and Logan (1984) 0.20 1.00
Reese and O’Neill (1987) 0.15 1.00
Toh et al. (1989) 0.25 1.00
Meigh and Wolshi (1979) 0.22 0.60
Horvath (1982) 0.20–0.30 0.50
Phoon (1993) defined αc as the ratio of the side shear resistance τmax to the undrained
shear strength cu. Understandably, the results of individual load tests show considerably
greater scatter than the site-averaged data. On the basis of the site-averaged data,
Kulhawy and Phoon (1993) proposed the following relations for drilled shafts in rock:
(6.6a)
(6.6b)
(6.6c)
Equation (6.6) can be rewritten in a general form as
(6.7)
This leads to a general expression for the side shear resistance

τmax=Ψ[paσc/2pa]−0.5
(6.8)
It is very important to note that the empirical relations given in Equations (6.6b) and
(6.6c) are bounds to site-averaged data, and do not necessarily represent bounds to
individual shaft behavior. The coefficient of determination (r2) is approximately 0.71 for
the site-averaged data, but is only 0.46 for the individual data, reflecting the much greater
variability of the individual test results (Seidel & Haberfield, 1995).
Fig. 6.2 Adhesion factor

αc(=τmax/0.5σc) versus normalized
shear strength for site-averaged data
(after Kulhawy & Phoon, 1993).
Fig. 6.3 Adhesion factor

αc(=τmax/0.5σc) versus normalized
shear strength for individual test data
(after Kulhawy & Phoon, 1993).
(c) Empirical relations considering roughness of shaft wall

The roughness of the shaft wall is an important factor controlling the development of side
shear resistance. Depending on the type of drilling technique and the hardness of the
rock, a drilled shaft will have a certain degree of roughness. Research has shown that the
benefits gained from increasing the roughness of a shaft wall can be quite significant,
both in terms of peak and residual shear resistance. Studies by Williams et al. (1980) and
others showed that smooth-sided shafts exhibit a brittle type of failure, while shafts
having an adequate roughness exhibit ductile failure. Williams and Pells (1981)
suggested that rough shafts generate a locked-in normal stress such that there is
practically no distinguishing difference between peak and residual side shear resistance.
Classifications have been developed so that roughness can be quantified. One such
classification proposed by Pells et al. (1980) is based on the size and frequency of
grooves in the shaft wall (see Table 6.4). Based on this classification, Rowe and
Armitage (1987b) proposed the following relation for shafts with different roughness:
τmax=0.45(σc)0.5 for shafts with roughness R1, R2 or R3
(6.9a)
τmax=0.60(σc)0.5 for shafts with roughness R4
(6.9b)
where both τmax and σc are in MPa.

Horvath et al. (1980) also developed a relation from model shaft behavior using
various roughness profiles. They found that as shaft profiles go from smooth to rough, the
roughness factor increases significantly, as does the peak side shear resistance. These
findings were confirmed in a later study by Horvath et al. (1983), and the following
equation was proposed for the roughness factor (RF):
(6.10)
where hm is the average roughness (asperity) height of the shaft; Lt is the total travel
length along the shaft wall profile; R is the nominal radius of the shaft; and L is the
nominal length of the shaft (see Fig. 6.4). Using Equation (6.10), the following relation
was developed between the side shear resistance and RF:
τmax= 0.8σc(RF)0.45
(6.11)
Kodikara et al. (1992) developed a rational model for predicting the relationship of τmax to
σc based on a specific definition of interface roughness, initial normal stress on the
interface and the stiffness of the rock during interface dilation. The parameters needed to
define interface roughness in the model are also shown in Figure 6.4. The model accounts
for variability in asperity height and angularity, assuming clean, triangular interface
discontinuities. Figure 6.5 shows the predicted adhesion factor, α(=τmax/σc), for
Melbourne Mudstone with the range of parameters and roughnesses as given in Table
6.5. The adhesion factor is presented as a function of Em/σc, σc/σn and the degree of
roughness, where Em is the elastic modulus of the rock mass and σn is the initial normal
stress on the shaft-rock interface. It can be seen that the adhesion factor is affected not
only by the interface roughness, but also by Em/σc and σc/σn.
Table 6.4 Roughness classes after Pells et al.

(1980).
Roughness Description
Class
R1 Straight, smooth-sided shaft, grooves or indentation less than 1.00 mm deep
R2 Grooves of depth 1–4 mm, width greater than 2 mm, at spacing 50 to 200 mm.
R3 Grooves of depth 4–10 mm, width greater than 5 mm, at spacing 50 to 200 mm.
R4 Grooves or undulations of depth greater than 10 mm, width greater than 10 mm,
at spacing 50 to 200 mm.
Fig. 6.4 Parameters for defining shaft

wall roughness (after Horvath et al.,
1980 and Kodikara et al., 1992).
Fig. 6.5 Simplified design charts for

adhesion factor α(=τmax/σc) for
Melbourne Mudstone (after Kodikara
et al., 1992).
Table 6.5 Definition of borehole roughness and

range of parameters for Melbourne Mudstone (after
Kodikara et al., 1992).
Range of values for shafts in Melbourne Mudstone
Parameter Smooth Medium Rough
im(degrees) 10–12 12–17 17–30
isd(degrees) 2–4 4–6 6–8
hm(mm) 1–4 4–20 20–80
hsd/hm 0.35
B(m) 0.5–2.0
σc(MPa) 0.5–10.0
σn(MPa) 50–500
Em(MPa) 50–500
Notes: 1) Refer to Figure 6.4 for the definitions of im, isd, hm and hsi
2) B=diameter of the shaft.
3) σc=unconfined compressive strength of the intact rock.
4) σn=initial normal stress on the shaft-rock interface.
5) Em=deformation modulus of the rock mass.
Seidel and Collingwood (2001) introduced a nondimensional factor called Shaft

Resistance Coefficient (SRC) to reflect the influence of shaft roughness and other factors
on the shaft side shear resistance. The SRC is defined as follows:
(6.12)
where hm is the mean roughness height (either assessed directly by estimation or

measurement, or computed as the product of asperity length, la, and the sine of the mean
asperity angle); B is the shaft diameter; ηc is the construction method reduction factor as
shown in Table 6.6; n is the ratio of rock mass modulus to the unconfined compressive
strength of the rock (Em/σc), known as the modulus ratio; and ν is the Poisson’s ratio of
the rock.
Using SRC, Seidel and Collingwood (2001) have created shaft resistance charts as
shown in Figures 6.6 and 6.7. These charts are based on results of a parametric study
using a computer program called ROCKET. To develop these charts, the intact rock
strength parameters were related to the unconfined compressive strength using the Hoek-
Brown strength criteria described in Chapter 4. Mohr-Coulomb strength parameters
adopted in the analyses were determined after the method of Hoek (1990) using the
unconfined compressive strength of the rock and appropriate values of parameters s and
m.
(d) Estimation of roughness height of shaft wall

Application of the empirical relations considering shaft wall roughness in design requires
estimation of likely shaft wall roughness height. A small number of studies have
produced actual roughness profiles which enable quantitative analysis. Detailed studies
have been carried out into shafts in Melbourne Mudstone (Williams, 1980; Holden, 1984;
Kodikara et al., 1992; Baycan, 1996). The results show that shaft wall roughness in this
low- to medium-strength argillaceous rock can vary considerably and appears to be
influenced by rock discontinuities, drilling techniques, and rate of advance. Shaft wall
roughness profiles in medium-strength shale were also recorded by Horvath et al. (1983),
but most of their shafts were artificially roughened by grooving. O’Neill & Hassan
(1994) and O’Neill et al., (1996) recorded measurements of roughness profiles of shafts
in clay shale, argillite and sandstone.
Table 6.6 Indicative construction method reduction
factor ηc (after Seidel & Collingwood, 2001).
Construction method ηc
Construction without drilling fluid
Best construction practice and high level of construction control 1.0
(e.g., shaft sidewalls free of smear and remoulded rock)
Poor construction practice or low-quality construction control (e.g., 0.3–0.9
smear or remoulded rock present on shaft sidewalls)
Construction under bentonite slurry
Best construction practice and high level of construction control 0.7–0.9
Poor construction practice or low-quality construction control 0.3–0.6
Construction under polymer slurry
Best construction practice and high level of construction control 0.9–1.0
Poor construction practice or low-quality construction control 0.8
Fig. 6.6 Adhesion factor α(=τmax/σc)

versus σc (after Seidel & Collingwood,
2001).
Fig. 6.7 Adhesion factor α(=τmax/σc)

versus SRC (after Seidel &
Collingwood, 2001).
Based on roughness heights back-calculated from load tests on shafts in rock, Seidel and
Collingwood (2001) developed the effective roughness height versus the unconfined
compressive strength plot as shown in Figure 6.8. The back-calculations were conducted
using Equation (6.12) and assuming ηc=1.0. In the case of a shaft for which the concrete-
rock interface is clean and unbounded, the roughness height back-calculated assuming
ηc=1.0 should provide a reasonable estimate of the roughness height magnitude.
However, if the shaft resistance is adversely influenced by construction procedures, the
roughness height would be underestimated if ηc is assumed to be 1.
Example 6.1
A drilled shaft of diameter 1.0 m is to be socketed 3.0 meters in rock. The rock properties
are as follows:
Unconfined compressive strength of intact rock, σc=15.0 MPa
Deformation modulus of intact rock, Er=10.6 GPa
RQD=76
Determine the side shear resistance.
Fig. 6.8 Effective roughness height

versus σc (after Seidel & Collingwood,
2001).
Solution:
Method of Kulhawy and Phoon (1993)—Equations (6.6) to (6.8)
Lower bound
τmax=1.0[paσc/2]0.5=1.0[0.1×15.0/2]0.5=0.87 MPa
Upper bound
τmax=3.0[paσc/2]0.5=3.0[0.1×15.0/2]0.5=2.60 MPa
Method of Seidel and Collingwood (2001)

From Figure 6.8, the mean roughness height hm=1.64 mm (lower bound) and 6.19 mm
(upper bound).
Using Equation (4.24), the rock mass modulus:
αE=0.0231(RQD)−1.32=0.297
Em=αEEr=0.297×10.6=3.15 GPa
The modulus ratio n=Em/σc=210.

The Poisson’s ratio of the rock is simply assumed to be ν=0.25.
Using ηc=1.0, SRC can be obtained from Equation (6.12) as:
From Figure 6.6, the adhesion factor a can be obtained as

α=0.102 (lower bound)
α=0.225 (upper bound)
So the side shear resistance can be obtained as

τmax=ασc=0.102×15.0=1.53 MPa (lower bound)
τmax=ασc=0.225×15.0=3.37 MPa (upper bound)
The results show that the shaft wall roughness (reflected by the roughness height) has
a great effect on the side shear resistance.
(e) Factors affecting side shear resistance

As stated above, the shaft wall roughness, which is an important factor controlling the
development of side shear resistance, has been studied extensively. Other factors such as
the discontinuities in the rock mass and the shaft geometry have also been studied by
some researchers. Williams et al. (1980) suggested that the existence of discontinuities in
the rock mass reduces the side shear resistance by reducing the normal stiffness of the
rock mass. They developed the following empirical relation that considers the effect of
discontinuities on the side shear resistance:
τmax=αwβwσc
(6.13)
where αw is a reduction factor reflecting the strength of the rock, as shown in Figure 6.9;
and βw is the ratio of side shear resistance of jointed rock mass to side shear resistance of
intact rock. βw is a function of modulus reduction factor, j, as shown in Figure 6.10, in
which
βw=f(j), j=Em/Er
(6.14)
where Em is the elastic modulus of the rock mass; and Er is the elastic modulus of the
intact rock. When the rock mass is such that the discontinuities are tightly closed and
seatns are infrequent, βw is essentially equal to 1.0. Comparing Equation (6.13) with
Equation (6.5), it can be seen that αwβw is just the adhesion factor, a, for β=1. Since αw is
derived from field test data, the effect of discontinuities is already included in αw. If αw is
multiplied by βw which is obtained from laboratory tests (Williams et al., 1980), the effect
of discontinuities will be considered twice. So Equation (6.13) may be too conservative.
Pabon and Nelson (1993) studied the effect of soft horizontal seams on the behavior of
laboratory model shafts. The study included four instrumented model shafts in
manufactured rock, three of which have soft seams. They concluded that a soft seam
significantly reduces the normal interface stresses generated in the rock layer overlying it.
Consequently the side shear resistance of shafts in rock with soft seams is much lower
than that of shafts in intact rock.
The effect of shaft geometry on side shear resistance was studied by Williams and Pells
(1981). They tested 15 shafts in Melbourne Mudstone, with diameters ranging from 335
mm to 1580 mm, and 27 shafts in Hawkesbury Sandstone, with diameters ranging from
64 mm to 710 mm. The results of these tests indicated that the shaft length, L, does not
have a discernible effect on the side shear resistance. They argued that the interface
dilation creates a locked-in normal stress with the result that the shear displacement
behavior exhibits virtually no peak or residual behavior. They also reported that the shaft
diameter has a negligible effect on the side shear resistance. On the other hand, tests by
Horvath et al. (1983) indicated that the side shear resistance decreases as the shaft
diameter increases. Williams and Pells (1981) explained this phenomenon by referring to
the theory of expansion of an infinite cylindrical cavity, which suggests that cylinders
with smaller diameters develop higher normal stresses for a given absolute value of
dilation. However, they offered no physical explanation why the shaft diameter does not
affect their own test results.
Fig. 6.9 Side shear resistance reduction

factor αw [Equation (6.13)] (after
Williams & Pells, 1981).
6.3.2 End bearing resistance
(a) End bearing behavior of drilled shafts

The typical bearing capacity failure modes for rock masses depend on discontinuity
spacing with respect to foundation width (or diameter), discontinuity orientation,
discontinuity condition (open or closed), and rock type. Table 6.7 illustrates typical
failure modes according to rock mass conditions (ASCE, 1996). Prototype failure modes
may actually consist of a combination of modes.
The failure modes shown in Table 6.7 are for foundations with the base at or close to
the ground surface. The depth of shaft embedment may change the end bearing failure
modes of drilled shafts. As shown in Figure 6.11, when the base of the shaft is at or close
to the ground surface, a wedge type of failure is developed and the shaft undergoes both
vertical settlement and rotation. When the depth of embedment is greater than twice the
diameter of the shaft, a punching type of failure occurs and a truncated conical plug of
fractured rock is formed below the base (Williams et al., 1980).
Fig. 6.10 Side shear resistance

reduction factor βw [Equation (6.13)]
(after Williams & Pells, 1981).
In a study by Johnston and Choi (1985), stereo photogrammetric techniques were used
to study the process of failure of a model pile socketed into simulated rock. As shown in
Figure 6.12, the study suggests that failure progresses from initial radial cracking to a fan
shaped wedge. These observations were compared to typical load displacement curve
where four points are identified as: 1) at the end of elastic deformation; 2) a little before
major yielding; 3) a little after major yielding; and 4) failure.
(b) End bearing resistance based on local experience and codes
Peck et al. (1974) suggested a correlation between the allowable bearing pressure and
RQD for footings supported on level surfaces in competent rock (Fig. 6.13). This
correlation can be used as a first crude step in determination of the end bearing resistance
of drilled shafts in rock. It need be noted that this correlation is intended only for
unweathered jointed rock where the discontinuities are generally tight. If the value of
allowable pressure exceeds the unconfined compressive strength of intact rock, the
allowable pressure is taken as the unconfined compressive strength.
In Hong Kong design practice, for large diameter drilled shafts in granitic and
volcanic rocks, the allowable end bearing resistance may be used as specified in Table
6.8. The presumptive end bearing resistance values range from 3.0 to 7.5 MPa, depending
on the rock category which is defined in terms of the rock decomposition grade, strength
and total core recovery.
Table 6.7 Typical bearing capacity failure modes
associated with various rock mass conditions (after
ASCE, 1996).
Rock mass conditions Failure
Joint Joint Illustration Mode
dip spacing
Brittle rock:
Local shear failure
caused by localized
brittle fracture
N/A s»B
Ductile rock:
General shear
failure along well
defined failure
surfaces
Open joints:
Compressive failure
of individual rock
columns. Near
70°<β vertical joint set(s)
<90°
s<B
Closed joints:
General shear
failure along well
defined failure
surfaces. Near
vertical joint(s)
s>B Open or closed

joints:
Failure initiated by
splitting leading to
general shear
failure. Near
vertical joint set(s)
20°<β s<B or s>B General shear

<70° if failure failure with
wedge can potential for failure
develop along joints.
along joints Moderately dipping
joint set(s)
Rock mass conditions Failure

Joint Joint Illustration Mode
dip spacing
Thick rigid upper layer:
Failure is initiated by
tensile failure caused by
flexure of the thick rigid
upper layer
Limiting
0°<β value of H
<20° with respect
to B Thin rigid upper layer:
Failure is initiated by
punching tensile failure
of the thin rigid upper
layer
General shear failure

with irregular failure
surface through rock
mass. Two or more
N/A s«B closely spaced joint sets
The Standard Specifications for Highway Bridges adopted by the American

Association of State Highway and Transportation Officials (AASHTO, 1989) also
provide presumptive allowable bearing pressures for spread footing foundations in rock
(see Table 6.9). These presumptive values can be used as a first crude step in
determination of the end bearing resistance of drilled shafts in rock.
(c) End bearing resistance from pressuremeter test results

The Canadian Foundation Engineering Manual (CGS, 1985) proposed a method for
determining the end bearing resistance of drilled shafts based on in situ pressuremeter test
results:
qmax=Kb(Pl−Po)+σo
(6.15)
Fig. 6.11 Typical failure mechanism

for end bearing shafts: (a) Base of shaft
bearing at ground surface; and (b)
Shaft with length/diameter>2 (after
Williams et al., 1980).
236
Fig. 6.12 Observed progressive failure

modes: (a) Typical load-displacement
curve; and (b) Failure modes
corresponding to the points in (a) (after
Johnston & Choi, 1985).
Fig. 6.13 Allowable bearing pressure

of jointed rock (after Peck et al., 1974).
Table 6.8 Presumed safe vertical bearing stress for
foundations on horizontal ground in Hong Kong
[simplified from PNAP 141 (BOO, 1990)].
Category Granitic and volcanic rock Presumed
bearing stress
(MPa)
1(a) Fresh to slightly decomposed strong rock of material weathering 7.5
grade II or better, with total core recovery>95 and minimum uniaxial
compressive strength of rock material σc not less than 50 MPa
(equivalent point load index strength PLI50a not less than 2 MPa)
1(b) Slightly to moderately decomposed moderately strong rock of 5.0
material weathering grade II or III or better, with total core
recovery>85% and minimum unconfined compressive strength of
rock material σc not less than 25 MPa (equivalent point load index
strength PLI50a not less than 1 MPa)
1(c) Moderately decomposed moderately strong to moderately weak rock 3.0
of material weathering grade III or IV or better, with total core
recovery>50%
a
Point load index strength PLI50 of rock quoted is equivalent value for 50-mm-diameter cores
(ISRM, 1979a).
where pl is the limit pressure as determined from pressuremeter tests in the zone
extending two shaft diameters above and below the shaft base; po is the at rest horizontal
238
stress in the rock at the elevation of the shaft base; σo is the total overburden stress at
elevation of the shaft base; and Kb is an empirical non-dimensional coefficient, which
depends on the depth and shaft diameter ratio as shown in Table 6.10.
(d) Empirical and Semi-Empirical Relations

Unlike the side shear resistance, numerous theories have been proposed for estimating the
end bearing resistance. According to Pells and Turner (1980), the theoretical approaches
fall into three categories:
1. Methods which assume rock failure to be plastic.
2. Methods which idealize the zone of failure beneath the base in a form which allows
either the brittleness strength ratio or the brittleness modulus ratio to be taken into
account.
3. Methods based on limiting the maximum stress beneath the loaded area to a value less
than required to initiate fracture. These methods assume essentially that once the
maximum strength is exceeded at any point in a brittle material, total collapse will
occur.
There is a significant variation in the end bearing resistance predicted from different
theories. For example, the predicted end bearing capacity of rock with an internal friction
angle ranges from 4.9σc using the incipient failure theory (Category 3) based
on the modified Griffith theory to 56σc using the classical plasticity theory (Category 1),
where σc is the unconfined compressive strength of intact rock (Poulos
Table 6.9 Presumptive allowable bearing pressures
for spread footing foundations, modified after Navy
(1982) (simplified from AASHTO, 1989).
Range of Allowable bearing pressure
σc (MPa) (MFa)
Type of bearing material Consistency in Ordinary Recommended
place range value for use
Massive crystalline igneous and Very hard, >250 6–10 8
metamorphic rock: granite, sound rock
diorite, basalt, gneiss,
thoroughly cemented
conglomerate (sound condition
allows minor cracks)
Foliated metamorphic rock: Hard sound rock 100–250 3–4 3.5
slate, schist (sound condition
allows minor cracks)
Sedimentary rock: hard Hard sound rock 50–100 1.5–2.5 2
cemented shales, siltstone,
sandstone, limestone without
cavaties
Weathered or broken bedrock Medium hard 25–50 0.8–1.2 1

of any kind except highly rock
argillaceous rock (shale)
Compaction shale or other Medium hard 25–50 0.8–1.2 1
highly argillaceous rock in rock
sound condition
Notes:
1. Variations of allowable bearing pressure for size, depth, and arrangement of
footings must be determined by analysis.
2. Presumptive values for allowable bearing pressures obtained from building codes
and charts developed by various agencies based on local experience with
satisfactory and unsatisfactory performance; usually the pressure that will limit
total and differential settlements to 1 inch. Presumptive values are not based on
thorough engineering analysis.
3. Allowable bearing pressure for rock is controlled by rock mass discontinuities, and
should not exceed the unconfined compressive strength.
Table 6.10 Kb as fimction of depth and shaft

diameter ratio (CGS, 1985).
Depth/Diameter 0 1 2 3 5 7
Kb 0.8 2.8 3.6 4.2 4.9 5.2
& Davis, 1980). Because of the wide variation of theoretical results, empirical and semi-
empirical relations have been developed. Since they are more commonly used than the
theoretical methods, only the empirical and semi-empirical relations are discussed in the
following.
Analogous to the side shear resistance, many attempts have been made to correlate the
end bearing capacity, qmax, to the unconfined compressive strength, σc, of intact rock.
Some of the suggested relations are:
Coates (1967): qmax=3.0σc
(6.16)
Rowe and Armitage (1987b): qmax=2.7σc
(6.17)
ARGEMA (1992): qmax=4.5σc ≤10 MPa
(6.18)
Findlay et al. (1997): qmax=(1−4.5)σc
(6.19)
The bearing capacity of foundations on rock is largely dependent on the strength of the
rock mass. Discontinuities can have a significant influence on the strength of the rock
mass depending on their orientation and the nature of material within discontinuities
(Pells & Turner, 1980). As a result, relations have been developed to account for the
influence of discontinuities in the rock mass. The Standard Specifications for Highway
Bridges adopted by the American Association of State Highway and Transportation
Officials (AASHTO, 1989) suggests that the end bearing capacity be estimated using the
following relationship:
qmax=Nmsσc
(6.20)
where Nms is a coefficient relating qmax to σc. The value of Nms is a function of rock mass
quality and rock type (Table 6.11), where rock mass quality, in essence, expresses the
degree of jointing and weathering. Rock mass quality has a much stronger effect on Nms
than rock type. For a given rock type, Nms for excellent rock mass quality is more than
250 times higher than Nms for poor quality. For a given rock mass quality, however, Nms
changes little with rock type. For example, for a rock mass of very good quality, the
values of Nms are 1.4, 1.6, 1.9, 2.0 and 2.3 respectively for rock types A, B, C, D and E
(see Table 6.11). It should be noted however that rock type is implicitly related to the
unconfined compressive strength. Equation (6.20) may thus represent a non-linear
relation between qmax and σc.
Although it is not explicitly mentioned in AASHTO (1989), Equation (6.20) and
coefficient Nms can be simply derived from the lower bound solution suggested by Carter
and Kulhawy (1988) (see Fig. 6.14):
qmax=[s0.5+(mbs0.5+s)0.5]σc
(6.21)
in which the expression in the brackets is simply the coefficient Nms in Equation (6.20);
and mb and s are the strength parameters for the Hoek-Brown strength criterion as
discussed in Chapter 4. Values of mb and s for the rock categories in Table 6.11 are
shown in Table 6.12. The values of Nms in Table 6.11 can be simply obtained by
inserting the corresponding values of mb and s from Table 6.12 in the expression in the
brackets of Equation (6.21).
Equation (6.21) does not consider the influence of the overburden soil and rock (i.e.,
overburden stress qs=0 is assumed). Zhang and Einstein (1998a) derived an expression
for the end bearing capacity that considers the influence of the overburden
Table 6.11 Values of Nms for estimating the end
bearing capacity of drilled shafts in broken or
jointed rock (after AASHTO, 1989).
Rock General RMR(1) Q(2) RQD(3) Nms(4)
Mask Description Rating Rating Rating
A B C D E
Quality
Excellent Intact rock with 100 500 95–100 3.8 4.3 5.0 5.2 6.1
joints spaced > 10
feet apart
Very Tightly interlocking, 85 100 90–95 1.4 1.6 1.9 2.0 2.3
Good undisturbed rock
with rough
unweathered
discontinuities
spaced 3 to 10 feet
apart
Good Fresh to slightly 65 10 75–90 0.28 0.32 0.38 0.40 0.46
weathered rock,
slightly disturbed
with discontinuities
spaced 3 to 10 feet
apart
Fair Rock with several 44 1 50–75 0.049 0.056 0.066 0.069 0.081
sets of moderately
weathered
discontinuities
spaced 1 to 3 feet
apart
Poor Rock with numerous 23 0.1 25–50 0.015 0.016 0.019 0.020 0.024
weathered
discontinuities
spaced 1 to 20
inches apart with
some gouge
Very Poor Rock with numerous 3 0.01 <25 Use qult for an equivalent soil
highly weathered
discontinuities
spaced<2 inches
apart
(1) Geomechanics rock mass rating (RMR) system (Bieniawski, 1988)—See Chapter 2
(2) Rock mass quality (Q) system (Barton et al., 1974)—See Chapter 2
(3) Range of RQD values provided for general guidance only; actual determination of rock
mass quality should be based on RMR or Q rating systems
(4) Value of Nms as function of rock type; refer to Table 2.8 for typical range of values of
σc for different rocks in each category
Table 6.12 Values of mb and s based on rock mass

classification (modified from Carter & Kulhawy,
1988).
Rock General RMR(1) Q(2) RQD(3) s mb
Mass Description Rating Rating Rating
A B C D E
Quality
Excellent Intact rock with 100 500 95–100 1 7 10 15 17 25
joints spaced
>10 feet apart
Very Tightly 85 100 90–95 0.1 3.5 5 7.5 8.5 12.5

Good interlocking,
undisturbed rock
with rough
unweath-ered
discontinuities
spaced 3 to 10
feet apart
Good Freshtoslightly 65 10 75–90 0.004 0.7 1 1.5 1.7 2.5
weathered rock,
slightly
disturbed with
discontinuities
spaced 3 to 10
feet apart
Fair Rock with 44 1 50–75 10−4 0.14 0.2 0.3 0.34 0.5
several sets of
moderately
weathered
discontinuities
spaced 1 to 3
feet apart
Poor Rock with 23 0.1 25–50 10−5 0.04 0.05 0.08 0.09 0.13
numerous
weathered
discontnuities
spaced 1 to 20
inches apart
with some
gouge
Very Rock with 3 0.01 <25 0 0.007 0.01 0.015 0.017 0.025
Poor numerous highly
weathered
discontinuities
spaced <2
inches apart
stress (see Fig. 6.15):
(6.22)
where
(6.23)
Fig. 6.14 Lower-bound solution for

bearing capacity (after Kulhawy &
Carter, 1992).
Fig. 6.15 Assumed failure mode of

rock mass below the shaft base (after
Zhang & Einstein, 1998a).
Kulhawy and Goodman (1980) presented the following relationship originally proposed
by Bishnoi (1968):
qmax=JcNcr
(6.24)
where J is a correction factor depending on normalized spacing of horizontal

discontinuities (spacing of horizontal discontinuities/shaft diameter) (see Fig. 6.16); c is
the cohesion of the rock mass; and Ncr is a modified bearing capacity factor, which is a
function of the friction angle of the rock mass and normalized spacing of vertical
discontinuities (spacing of vertical discontinuities/shaft diameter) (see Fig. 6.17).
Fig. 6.16 Correction factor for

discontinuity spacing (after Kulhawy
& Goodman, 1980).
Fig. 6.17 Bearing capacity factor for

open discontinuities (after Kulhawy &
Goodman, 1980).
Table 6.13 Suggested design values of strength
parameters c and (from Kulhawy Goodman,
1987).
Rock mass properties
RQD (%) Unconfined Compressive strength Cohesion c
Angle of friction
0–70 0.33σc 0.1σc 30°
70–100 0.33σc–0.8σc 0.1σc 30°–60°
σc=unconfined compressive strength of intact rock
As indicated in the preceding text, the strength parameters c and are rock mass properties.
Kulhawy and Goodman (1987) provided a table relating the rock mass properties c and
to intact rock properties and RQD (Table 6.13). The correction factor J considers the
effect of horizontal discontinuities and the variation of Jwith the discontinuity spacing is
shown in Figure 6.16, where H is the spacing of horizontal discontinuities. For the value
of Ncr the authors considered the discontinuities being either open or closed. According to
Goodman (1980), the presence of open discontinuities would allow failure to occur by
splitting (because the discontinuities are open, there is no confining pressure and failure
is likely to occur by uniaxial compression of the rock columns), and this mode of failure
needs to be included in the calculation of the end bearing capacity. Several charts are
given by Kulhawy and Goodman (1980), following the method of Bishnoi (1968), to
determine Ncr for both open and closed discontinuities. Figure 6.17 shows Ncr for open
discontinuities.
The Canadian Foundation Engineering Manual (CGS, 1985) proposed that the end
bearing pressure be calculated using the following equation:
qmax=3σcKspD
(6.25)
where Ksp=[3+s/B]/[10(1+300g/s)0.5] is an empirical factor; s is the spacing of the

discontinuities; B is the shaft width or diameter; g is the aperture of the discontinuities;
D=1+0.4(L/B)≤3.4 is the depth factor; L is the shaft length. In general the method will
apply only if s/B ratios lie between 0.05 to 2.0 and the values of g/s between 0 and 0.02
(CGS, 1985).
Zhang and Einstein (1998a) developed a database of 39 shaft load tests about the
ultimate end bearing capacity (see Table 6.14). This database represents rocks of
relatively low strength. Table 6.14 lists, in addition to shaft dimensions, the unconfined
compressive strength of intact rock σc, the end bearing capacity qmax, and the end bearing
capacity factor Nc(=qmax/σc). The ratio of the shaft base displacement sb at qmax to the
shaft diameter B is also included in Table 6.14. A number of issues that need be
considered when studying the relationship between the end bearing capacity and the
unconfined compressive strength of intact rock are as follows:
1. Different interpretations of the load test data will give different capacities. For the test
shafts in Table 6.14, different interpretation methods were used. For example, Goeke
and Hustad (1979) took the load at plunging failure as the ultimate capacity of the
shaft (plunging failure is defined as the point at which additional load cannot be
applied to the shaft without experiencing continuous movement), while Jubenville and
Hepworth (1981) defined the ultimate capacity of the shaft as the load at which the
shaft head displacement reached 10% of the shaft diameter. Therefore, some
uncertainties and variabilities are likely to be included in the database. However, the
general trend reflected by the database will be useful.
2. The unconfined compressive strength is a property of the intact rock, not of the rock
mass. Clearly rock mass discontinuities must affect the end bearing
Table 6.14 Summary of database of shaft load tests
(Zhang and Einstein, 1998a).
No. Rock Diameter Depth σc qmax Nc=qmax/σc Sb/Ba Reference
description B (mm) to (MPa) (MPa) (%)
base L
(m)
1 Mudstone, 670 6 1.09 6.88 6.31 7.0 Wilson (1976)
weak, clayey
cretaceous
2 Clayshale, 762 8.8 0.81 4.69 5.79 6.2 Goeke and
with Hustad(1979)
occational thin
limestone
seams
3 Shale, thinly 457 13.7 3.82 10.8 2.83 >10.0 Hummert and
bedded with Cooling (1988)
thin sandstone
layers
4 Shale, 305 2.4 1.08 3.66 3.39 10.0 Jubenville and
unweathered Hepworth(1981)
5 Gypsumb 1064 4.20 2.1 6.51 3.1 15– Leung and Ko
20 (1993)
20 (1993)
20 (1993)
20 (1993)
9 Gypsumb 1064 4.20 8.5 23 2.7 15– Leung and Ko
20 (1993)
20 (1993)
11 Tillc 762 ** 0.7 4 5.71 ~1.3 Orpwoodetal.
(1989)
12 Tillc 762 ** 0.81 4.15 5.12 ~4.6 Orpwood et al.
(1989)
13 Tillc 762 ** 1 5.5 5.5 ~1.4 Orpwood et al.
(1989)
14 Diabase, 615 12.2 0.52 2.65 5.1 >4.0 Webb (1976)
highly
weathered
15 Hardpan (hard 1281 18.3 1.38 5.84 4.23 ~4.0 Baker (1985)
bearing till)c
16 Tillc 1920 20.7 0.57 2.29 4.04 ~1.9 Baker (1985)
17 Hardpan (hard 762 18.3 1.11 4.79 4.33 ~7.3 Baker (1985)
bearing till)c
18 Sandstone, 610 15.6 8.36 10.1 1.21 >1.7 Glos and Briggs
horizontally (1983)
bedded,
shaley,
RQD=74%
19 Sandstone, 610 16.9 9.26 13.1 1.41 >1.7 Glos and Briggs
horizontally (1983)
bedded,
shaley, with
some coal
stringers,
RQD=88%
20 Mudstone, 300 2.01 0.65 6.4 9.8 6.4 Williams (1980)
highly
weathered
21 Mudstone, 300 1 0.67 7 10.5 5.7 Williams (1980)
highly
weathered
moderately
weathered
moderately
weathered
24 Mudstone, 1000 15.5 2.45 7 2.9 0.6 Williams (1980)
moderately
weathered
moderately
weathered
moderately
weathered
27 Mudstone, 1000 3 1.4 7.1 5 10.9 Williams (1980)
moderately
weathered
No. Rock Diameter Depth σc qmax Nc= Sb/Ba Reference

description B (mm) to (MPa) (MPa) qmax/σc (%)
base L
(m)
28 Shale ** ** 34 28 0.82 ** Thorne (1980)d
29 Sandstone ** ** 12.5 14 1.12 ** Thorne (1980)d
30 Sandstone, fresh, ** ** 27.5 50 1.82 ** Thorne (1980)d
defect free
31 Shale, occational ** ** 55 27.8 0.51 ** Thorne (1980)d
recemented
moisture
fractures and
thin mud seams,
intact core
lengths 75 to
250 mm
32 Clayshale 740 7.24 1.42 5.68 4 ~8.8 Aurora and
Reese (1977)
33 Clayshale 790 7.29 1.42 5.11 3.6 ~8.9 Aurora and
Reese (1977)
Reese (1977)
Reese (1977)
36 Siltstone, 705 7.3 9 13.1 1.46 ~12.0 Radhakrishnan
medium hard, and Leung
fragmented (1989)
37 Marl, intact, 1200 18.5 0.9 5.3 5.89 ** Carrubba
RQD=100% (1997)
38 Diabase Breccia, 1200 19 15.0 8.9 0.59 ** Carrubba
highly fractured, (1997)
RQD=10%
39 Limestone, 1200 13.5 2.5 8.9 3.56 ** Carrubba
intact, (1997)
RQD=100%
a
Sb is the shaft base displacement at qmax.
b
Gypsum mixed with cement is used as pseudo-rock in centrifuge tests. The and depths are the
equivalent prototype dimensions corresponding to 40 g in the centrifuge tests. The equivalent
prototype depths to the shaft base range 4.04 m to 4.35 m with an average of 4.20 m.
c
Till is not a rock. It is used here because its σc is comparable to that of some rocks.
d
These tests were not conducted by Thorne (1980). He only reported the data other references
capacity. Unfortunately, relevant information on this factor is unavailable for

most of the cases in Table 6.14.
3. The conditions below the base of the shaft also influence the end bearing capacity. If
the base of the drilled hole cannot be cleaned, little or no end bearing support will be
developed. For all the test shafts in Table 6.14, the base of the drilled hole was
cleaned.
4. Different methods are used to separate the side shear resistance from the end bearing
capacity in load tests.
5. Clearly it would be interesting to have a relatively narrowly defined shaft base
displacement which one can associate with the end bearing capacity. However, the
values of sb/B in Table 6.14 indicate that the base displacement at qmax ranges from
0.6 to 20% of the shaft diameter, i.e., 6 to 210 mm. It is thus difficult to say at this
point what typical base displacements at qmax are. [For comparison, the displacement
at ultimate side shear resistance is smaller; examination of more than 50 load-
displacement curves for large-diameter drilled shafts showed that an average
displacement of only 5 mm was necessary to reach initial failure of side shear
resistance (Horvath et al., 1983)].
Fig. 6.18 qmax versus σc (after Zhang &

Einstein, 1998a).
All the load test data in Table 6.14 are plotted in Figure 6.18. A log-log plot is used. It
can be seen that there is a strong relation between qmax and σc. Using linear regression, the
relationship between qmax and σc is as follows:
qmax=4.83(σc)0.51
(6.26)
The coefficient of determination, r2, is 0.81.
Example 6.2
A drilled shaft of diameter 1.0 m is to be socketed 3.0 meters in siltstone. The rock
properties are as follows:
The rock mass is heavily jointed and the average discontinuity spacing near
the base of the shaft is 0.5 m
The discontinuities are moderately weathered and filled with debris with
thickness of 3 mm
RQD=45
Determine the end bearing resistance.
Solution:
Method of AASHTO (1989)—Equation (6.20)
From Table 2.8, the rock is classified as Type B.
From Table 6.11, the rock quality is classified as Fair and the value of Nms is 0.056.
Using Equation (6.20), the end bearing resistance can be obtained as
qmax=Nmsσc=0.056×15.0=0.84 MPa
Method of Zhang & Einstein (1998a)—Equations (6.22) & (6.23)

From Table 2.8, the rock is classified as Type B.
From Table 6.12, the rock quality is classified as Fair and the values of s and mb are
respectively 10−4 and 0.2.
Assuming that the effective unit weight of the rock mass is 13.0 kN/m3 and ignoring
the weight of the soil above the rock, the end bearing resistance can be obtained from
Equations (6.22) and (6.23) as
Method of CGS (1985)—Equation (6.25)

Empirical factor Ksp=[3+s/B]/[10(1+300g/s)0.5]=(3+0.5/1.0)/[10(1+300×0.003/0.5)0.5]
=0.21
Depth factor D=1+0.4(L/B)=1+0.4(3.0/1.0)=2.2.
The end bearing resistance can be calculated from Equation (6.25) as
Method of Zhang and Einstein (1998)—Equation (6.26)

The end bearing resistance can be simply calculated from Equation (6.26) as
The results clearly show the wide range of the estimated end bearing capacity from
different methods. It is therefore important not to rely on a single method when
estimating the end bearing capacity.
6.4 CAPACITY OF DMLLED SHAFT GROUPS
In many cases, drilled shaft foundations will consist not of a single drilled shaft, but of a
group of drilled shafts. The drilled shafts in a group and the soil/rock between them
interact in a very complex fashion, and the axial capacity of the group may not be equal
to the axial capacity of a single isolated drilled shaft multiplied by the number of shafts.
One way to account for the interaction is to use the group efficiency factor η, which is
expressed as:
(6.27)
where QuG is the ultimate axial load of a drilled shaft group; N is the number of drilled
shafts in the group; and Qu is the ultimate axial load of a single isolated drilled shaft,
which can be determined using the methods described in Section 6.3. The group
efficiency for axial load capacity depends on many factors, including the following:
•The number, length, diameter, arrangement and spacing of the drilled shafts.
•The load transfer mode (side shear versus end bearing).
•The elapsed time since the drilled shafts were installed.
•The rock type.
Katzenbach et al. (1998) studied the group efficiency of a large drilled shaft group in
rock. For the 300 m high Commerzbank tower in Frankfurt am Main, 111 drilled shafts
are used to transfer the building load through the relatively weak Frankfurt Clay to the
stiffer underlying Frankfurt Limestone. Of the 111 drilled shafts, 30 were instrumented
and monitored during the 2-year construction period. The measurements give a detailed
view into the interaction between the drilled shafts in the group. Figure 6.19 shows the
variation of the group efficiency factor with the shaft head settlement. At service loads of
the building the value of the group efficiency factor is about 60%.
When drilled shafts are closely spaced, the shafts in a group may tend to form a
“group block” that behaves like a giant, short shaft (see Fig. 6.20). In this case, the
bearing capacity of the drilled shaft group can be obtained in a similar fashion to that for
a single isolated drilled shaft, by means of Equation (6.2), but now taking the shaft base
area as the block base area and the shaft side surface area as the block surface area. It
should be noted that the deformation required to mobilize the base capacity of the block
will be larger than that required for a single isolated shaft.
6.5 UPLIFT CAPACITY
In many cases, drilled shafts in rock may be required to resist uplift forces. Examples are
drilled shaft foundations for structures subjected to large overturning moments such as
tall chimneys, transmission lines, and highway sign posts. Drilled shafts through
expansive soils and socketed into rock may also subject to uplift forces due to the
swelling of the soil.
Drilled shafts can be designed to resist uplift forces either by enlarging or belling the
base, or by developing sufficient side shear resistance. Belling the base of a shaft is
common in soils, but this can be an expensive and difficult operation in rock. Moreover,
since large side shear resistance can be developed in drilled shafts socketed into rock, it is
usually more economical to deepen the socket than to construct a shorter, belled socket.
For drilled shafts subject to uplift forces, it is important to check the structural
capacity of the shaft. This can be done using the methods presented in Section 6.1. The
ultimate uplift resistance of a straight-sided drilled shaft related to rock can be
determined by
Quu=πBLτmax+Ws
(6.28)
where Quu is the ultimate uplift resistance; L and B are respectively the length and
diameter of the shaft; τmax is the average side shear resistance along the shaft; and Ws is
the weight of the shaft.
Fig. 6.19 Variation of group efficiency

factor with shaft head settlement (after
Katzenbach et al., 1998).
Fig. 6.20 Treating the drilled shaft

group as a group block.
Uplift loading does not produce the same stress conditions in the shaft or rock mass as
those produced by compression loading. Compression loading compresses the shaft,
causing outward radial straining in the concrete (positive Poisson effect), which results in
higher frictional stresses at the interface with the rock mass; simultaneously it adds total
vertical stress to the rock mass around the shaft through the process of load transfer,
which consequently adds strength to rock masses that drain during loading. Uplift
loading, however, produces radial contraction of the concrete (negative Poisson effect)
and reduces the total vertical stresses in the rock mass around the shaft. Because of the
different stress conditions, the average side shear resistance for uplift loading should
usually be lower than that for compression loading.
Fig. 6.21 Measured side shear

resistance from compression tests and
pull-out tests.
Figure 6.21 shows the variation of measured side shear resistance with the unconfined
compressive strength of intact rock respectively from the compression load tests and the
pull-out load tests. The data are collected from the published literature. We can see that
the measured side shear resistances from the pull-out load tests are about the same as or
even higher than those from the compression load tests. One of the reasons for this might
be that the pull-out test shafts have rougher wall surfaces than the compression test
shafts. However, we are not sure about this at this point since no information on the wall
roughness is available for most of the test shafts shown in Figure 6.21.
For preliminary design, the side shear resistance for uplift loading can be simply taken
to be the same as that for compression loading and estimated using the methods presented
in Section 6.3.1.
Where vertical drilled shafts are arranged in closely-spaced groups the uplift resistance
of the complete group may not be equal to the sum of the resistance of the individual
shafts. This is because, at ultimate-load conditions, the block of rock enclosed by the
shafts may be lifted. The uplift resistance of the block of rock may be determined by (see
Fig. 6.20)
(6.29)
where QuuG is the total ultimate uplift resistance of the shaft group; B1 and B2 are
respectively the overall length and width of the group (see Fig. 6.20); and WB is the
combined weight of the block of rock enclosed by the shaft group plus the weight of the
shafts.
7
Axial deformation of drilled shafts in rock
7.1 INTRODUCTION
Predicting the axial load-displacement response of drilled shafts is in some cases as

important as, or possibly more critical than, predicting the ultimate bearing capacity.
Many methods are available for predicting the axial displacement of drilled shafts in
rock. While the most reliable means for predicting the axial displacement of drilled shafts
is probably to carry out an axial loading test of the prototype shaft (which will be
discussed in Chapter 12), theoretical analyses may also be usefully employed. The main
three theoretical methods used to predict the axial load-displacement response of drilled
shafts in rock are the load-transfer (t-z) method, the continuum approach and the finite
element method.
The general load-displacement curve for a drilled shaft under axial loading can be
simply illustrated in Figure 7.1. The whole curve can be described in three stages:
1. As load is first applied to the head of the shaft, a small amount of displacement occurs
which induces the mobilization of side shear resistance from head to base. During this
initial period, the shaft behaves essentially in a linear manner, and the displacement
can be computed using the theory of elasticity. This linear behavior is illustrated in
Figure 7.1 as the line OA. The side shear stress along the shaft is smaller than the
ultimate side shear resistance (Fig. 7.2a).
2. As load is increased to point A in Figure 7.1, the shear stress at some point along the
interface will reach the ultimate side shear resistance (Fig. 7.2b), and the shaft-rock
‘bound’ will begin to rupture and relative displacement (slip) will occur between the
shaft and the surrounding rock. As the loading is increased further (beyond point A),
this process will continue along the shaft, more of the shaft will slip, and a greater
proportion of the applied load will be transferred to the end of the shaft (Fig. 7.2c). If
loading is continued, eventually the side shear stress everywhere will reach the
ultimate side shear resistance and the entire shaft will slip (point B in Fig. 7.1).
3. Beyond point B, a greater proportion of the total axial load will be transmitted directly
to the end of the shaft. When both side shear resistance and end bearing resistance are
fully mobilized (point C), any increase of load may produce significant displacement.
This indicates that the ultimate bearing capacity of the drilled shaft has been reached.
Axial deformation of drilled shafts in rock 259
Fig. 7.1 Generalized load-displacement

curves for drilled shafts under
compressive loading.
7.2 LOAD-TRANSFER (t-z CURVE) METHOD
The load-transfer method models the reaction of soil/rock surrounding the shaft using
localized springs: a series of springs along the shaft (the t-z or τ-w curves) and a spring at
the tip or bottom of the shaft (the q-w curve). τ is the local load transfer or side shear
resistance developed at displacement w, q is the base resistance developed at
displacement w, and w is the displacement of the shaft at the location of a spring. The
physical drilled shaft is also represented by a number of blocks connected by springs to
indicate that there will be compression of the drilled shaft due to the applied compressive
load. The mechanical model is shown in Figure 7.3. The displacement of the shaft at any
depth z can be expressed by the following differential equation:
(7.1)
where Ep is the composite Young’s modulus of the shaft (considering the contribution of
both concrete and reinforcing steel); A and B are respectively the cross-sectional area and
diameter of the shaft; w is the displacement of the shaft at depth z; and τ is the side shear
resistance developed at displacement w at depth z.
Equation (7.1) can be solved analytically or numerically depending on the τ-w and q-w
curves (linear or nonlinear), which is discussed in the sections below.
7.2.1 Linear analysis

For linear analysis, the relationship between τ and w at any depth z and that between q
and w are assumed to be linear, i.e.,
Fig. 7.2 Shear stress at different values

of applied load (QA is the applied load
corresponding to point A in Fig. 7.1).
(7.2a)
(7.2b)
where ks and kb are spring constants respectively of the side springs and the base spring.
Substitution Equation (7.2a) into Equation (7.1) gives
(7.3)
where
Fig. 7.3 Load-transfer (t-z curve)

model of axially loaded drilled shaft.
(7.4)
The general solution to Equation (7.3) is

(7.5)
where C1 and C2 are integration constants.

The axial force at any depth is proportional to the first derivative of the displacement
with respect to depth:
(7.6)
If a load Qt is applied at the top of the shaft (z=0) and the force transferred to the base of
the shaft (z=L) is Qb, we have, from Equation (7.6),
(7.7a)
(7.7b)
From Equations (7.2b) and (7.5), We have
(7.8)
Solving Equations (7.7) and (7.8), constants C1 and C2 can be obtained as
(7.9a)
(7.9a)
The displacement at the top of the shaft (z=0) is then obtained from Equations (7.5) and
(7.9) as
7.2.2 Nonlinear analysis

In general, the τ-w and q-w curves are nonlinear. In this case, a convenient way to solve
differential Equation (7.1) is to use the finite difference method (Desai & Christian,
1977). Computer programs can be easily written to do the computations. The main issue
for the nonlinear analysis is the determination of the τ-w and q-w curves.
There are several techniques for determining the load transfer curves in soils
(Vijayvergiya, 1977; Kraft et al., 1981; Castelli et al., 1992) and rock masses (Baguelin et
al., 1982; O’Neill & Hassan, 1994). However, research has not advanced to the point that
the load transfer curves (τ-w and q-w curves) can be determined for all conditions with
confidence (O’Neill & Reese, 1999). Construction practices and the particular response
of a given formation to drilling and concreting will affect the load transfer curves. For
major projects, therefore, it is advisable to measure the load transfer curves using full-
scale loading tests of instrumented shafts. Chapter 12 will show how to obtain the
experimental load transfer curves from the results of an axial loading test of an
instrumented shaft.
Based on measured load displacement curves, Carrubba (1997) conducted numerical
analyses to evaluate the side shear resistance and the end bearing capacity and obtained
the load transfer curves for five rock-socketed shafts. The model is based on a hyperbolic
transfer function approach and solves the equilibrium of the shaft by
means of finite element discretization. The interaction at the shaft-soil and shaft-rock
interfaces is described by the following function
(7.11)
where f(z) is the mobilized resistance along a shaft portion (τ) or at the shaft base (q); and
w(z) is the corresponding displacement (see Fig. 7.3). In the transfer function, parameters
a and b represent the reciprocals of initial slope and limit strength, respectively:
(7.12a)
(7.12b)
where flim is the end bearing capacity (qmax) in rock or the side shear resistance in soil or
rock (τmax).
Numerical analyses are carried out by selecting three transfer functions for each shaft:
one representative of overall friction in soil, one for overall friction in rock, and the last
one for end bearing resistance in rock. The friction transfer functions in soils, once
selected, are maintained constant throughout the analyses. Transfer function parameters
for rock, both along the shaft and at the base, are first estimated and then modified with
an iterative process until the actual load displacement curve is reproduced. Figure 7.4
shows the comparison between the test results and the numerical simulations for the shaft
in marl. Since the side and base strengths are not mobilized at the same time and the
numerical model used cannot simulate this event, two different ideal shaft behaviors are
examined. The first neglects the base reaction; the second takes into account the
contemporary mobilization of side and base resistances from the beginning of the test.
The rock properties and the transfer function parameters obtained for the five rock-
socketed shafts are shown in Table 7.1.
O’Neill and Hassan (1994) proposed an interim criterion for a hyperbolic τ-w curve in
most types of rock until better solutions become accepted:
(7.13)
where B is the diameter of the shaft; and Em is the deformation modulus of the rock mass.
This model is based on the fact that the interface asperity pattern is regular and the
asperities are rigid, even though in most cases the interface asperity pattern is not regular,
some degree of smear exists, and asperities are deformable, which results in ductile,
progressive failure among asperities. Equation (7.13) is a special form of Equation (7.11)
with a=2.5B/Em.
Fig. 7.4 Comparison between test

results and numerical simulations for
the drilled shaft in Marl. Curve a
neglects base reaction; curve b takes
into account cotemporary mobilization
of side and base resistances (after
Carrubba, 1997).
Table 7.1 Rock properties and transfer function
parameters (Carrubba, 1997).
Shaft side in rock Base in rock
Rock type σc RQD Em 1/b 1/a 1/b 1/a
(MPa) (%) (MPa) (MPa) (MN/m3) (MPa) (MN/m3)
Marl 0.90 100 200a 0.14 100 5.30 220
b
Diabasic 15.00 10 200 0.49 70 8.90 300
Breccia
Gypsum 6.00 60 2,000a 0.47 200 – –
a
Diabase 40.00 50 10,000 1.20 500 – –
b
Limestone 2.50 100 500 0.40 500 8.90 3,000
a
From compression tests on specimens
b
From plate bearing tests
The q-w curve is usually assumed to have an initial elastic response given by
where Eb and νb are respectively the deformation modulus and Poisson’s ratio of the rock
below the shaft base. Nonlinear response is usually assumed to initiate between 1/3 and
1/2 of qmax. This response can be simply modeled using an equation similar to Equation
(7.13).
7.3 CONTINUUM APPROACH
The continuum approach assumes the soil/rock to be a continuum. Mattes and Poulos
(1969) are among the first to investigate the load-displacement behavior of rock-socketed
shafts by integration of Mindlin’s equations. Carter and Kulhawy (1988) provide a set of
approximate analytical solutions to predict the load-displacement response of drilled
shafts in rock by modifying the solutions of Randolph and Wroth (1978) for piles in soil.
The majority of the theoretical continuum solutions for predicting the displacement of
drilled shafts in rock, however, have been developed using finite element analyses (e.g.,
Osterberg & Gill, 1973; Pells & Turner, 1979; Donald et al., 1980; Rowe & Armitage,
1987a). Most of the techniques proposed for calculating the vertical displacements of
drilled shafts in rock are based on the theory of elasticity. It has been usual to assume that
the drilled shaft is essentially an elastic inclusion within the surrounding rock mass and
that no slip occurs at the interface between the shaft and the rock mass, although the
solutions of Rowe and Armitage (1987a) and Carter and Kulhawy (1988) can consider
the possibility of slip.
7.3.1 Linear continuum approach
(a) Solutions based on finite element results

As stated in Chapter 6, axially loaded drilled shafts in rock are designed to transfer
structural loads in one of the following three ways (CGS, 1985):
1. Through side shear only;
2. Through end bearing only;
3. Through the combination of side shear and end bearing.
The following presents the elastic solutions based on the finite element results for
estimating the axial deformation of the above three types of shafts.
Side shear only shaft

Based on finite element analysis, Pells and Turner (1979) presented the following general
equation for calculating the axial deformation of side shear only shafts in a single elastic
half space:
(7.15)
where wt is the axial deformation of the shaft at the rock surface; Qt is the applied load at
the top of the shaft; Em is the deformation modulus of the rock mass; B is the diameter of
the shaft; and I is the axial deformation influence factor given in Figure 7.5. The values
of I given in Figure 7.5 have been calculated for a Poisson’s ratio of 0.25. It has been
found that variations in the Poisson’s ratio in the range 0.1–0.3 for the rock mass and
0.15–0.3 for the concrete have little effect on the influence factors.
The values of the influence factor shown in Figure 7.5 are for drilled shafts that are
fully bonded from the rock surface. In many cases, the drilled shaft is recessed by casing
the upper part of the drilled hole or for conditions where the shaft passes through a layer
of soil or weathered rock where little or no side shear resistance will be developed.
Recessment of the shaft will result in a decrease in axial deformation of the shaft at the
head of the socket. This reduction can be expressed in terms of a reduction factor RF
such that the axial deformation of the shaft at the ground surface is given by
(7.16)
Fig. 7.5 Axial deformation influence

factors for side shear only drilled
shafts (after Pells & Turner, 1979).
where Qt is the applied load at the top of the shaft; D and Bl are respectively the length
and diameter of the recessed shaft; Ep is the composite Young’s modulus of the shaft
(considering contributions of both concrete and reinforcing steel); RF is a reduction
factor for the effect of recessment; B is the diameter of the socketed shaft; Em is the
deformation modulus of the rock mass; and I is the influence factor for shaft with no
recessment (see Fig. 7.5). The first portion of Equation (7.16) simply represents the
elastic compression of the shaft over the length D. The second portion of Equation (7.16)
gives the axial deformation of the socketed portion of the shaft. The reduction factor RF
is given in Figure 7.6 for a range of situations.
Fig. 7.6 Reduction factors for

calculation of axial deformation of
recessed drilled shafts (after Pells &
Turner, 1979).
End bearing only shaft
An end bearing only shaft can be considered a shaft that is wholly recessed (See Fig. 7.7).
The axial deformation of an end bearing only shaft at the ground surface consists of the
elastic compression of the shaft and the axial deformation of the shaft base:
(7.17)
where Qt is the applied load at the top of the shaft; D and Bl are respectively the length
and diameter of the shaft; Ep is the composite Young’s modulus of the shaft (considering
contributions of both concrete and reinforcing steel); Em and νm are respectively the
deformation modulus and Poisson’s ratio of the rock mass; Cd is the shape and rigidity
factor equal to 0.85 for a flexible footing and 0.79 for a rigid footing; and RF′ is a
reduction factor for an end bearing only shaft as shown in Figure 7.7.
The axial deformation of the shaft base is calculated in a similar manner to that of a
footing on the surface. However, because the rock mass below the base of the shaft is
more confined than surface rock mass, the axial deformation of the shaft base will be
smaller than that of a footing at the surface. The effect of this confinement if accounted
for by applying the reduction factor RF′ to the deformation equation as shown in
Equation (7.17). The value of the reduction factor depends on the ratio of the shaft length
D to the shaft diameter B1, and the relative stiffness of the shaft and the rock mass. Figure
7.7 shows the values of the reduction factor RF′ obtained by Pells and Turner (1979).
Side shear and end bearing shaft

For side shear and ending bearing shafts, the axial deformation at the rock surface can be
calculated using Equation (7.15). Considering the interaction between the side shear and
end bearing, the influence factors given in Figure 7.8 should be used. These factors have
been developed for elastic behavior without slip along the side walls by Rowe and
Armitage (1987a).
Fig. 7.7 Reduction factors for

calculation of axial deformation of end
bearing only drilled shafts (after Pells
& Turner, 1979).
Comparison of Figure 7.8(a) (for Eb/Em=1) with Figure 7.5 shows that the influence
factor for a side shear and end bearing shaft is smaller than that for a side shear only
shaft, which demonstrates that a shaft with both side shear and end bearing will settle less
than a shaft with side shear only. Figure 7.9 shows the percentage of the load carried in
the end bearing.
(b) Analytical solutions of Carter and Kulhawy (1988)

Carter and Kulhawy (1988) provide a set of approximate analytical solutions to predict
the load-displacement response of drilled shafts in rock. Two layers of rock mass as
shown in Figure 7.10 are considered in the solutions. The solutions are for a shaft without
slip or with full slip. The following presents the solution for a shaft without slip while the
solution for a shaft with full slip will be presented in Section 7.3.2.
Under an applied axial load, the displacements in the rock mass are predominantly
vertical, and the load is transferred from the shaft to the rock mass by vertical shear
stresses acting on the cylindrical interface, with little change in vertical normal stress in
the rock mass (except near the base of the shaft). The pattern of deformation around the
shaft may be visualized as an infinite number of concentric cylinders sliding inside each
other (Randolph & Wroth, 1978). Randolph and Wroth (1978) have shown that, for this
type of behavior, the displacement of the shaft w may be described adequately in terms of
hyperbolic sine and cosine functions of depth z below the surface, as given below:
w=A1 sinh(µz)+A2 cosh(µz)
(7.18)
in which, A1 and A2 are constants which can be determined from the boundary conditions
of the problem. The constant µ is given by
(7.19)
where ζ=ln[2.5(1−νm)L/R]; R=B/2 is the radius of the shaft; λ=Ep/Gm; Ep is the Young’s
modulus of the shaft; Gm=Em/[2(1+νm)] is the shear modulus of the rock mass
surrounding the shaft; and Em and νm are respectively the deformation modulus and
Poisson’s ratio of the rock mass surrounding the shaft.
For side shear and end bearing shafts as shown in Figure 7.10(a), the shaft base can be
approximated as a punch acting on the surface of an elastic half-space with Young’s
modulus Eb and Poisson’s ratio νb. Using the standard solutions for the displacement of a
rigid punch resting on an elastic half-space as the boundary condition at the base of the
shaft, the elastic displacement at the head of the shaft can be obtained by (Randolph &
Wroth, 1978):
(7.20)
Fig. 7.8 Axial deformation influence

factors for side shear and end bearing
drilled shafts (after Rowe & Armitage,
1987a).
Fig. 7.9 Load distribution curves for

side shear and end bearing drilled
shafts (after Rowe & Armitage,
1987a).
where ξ=Gb/Gm; Gb=Eb/[2(1+νb)] is the shear modulus of the rock mass below the shaft
base; and Eb and νb are respectively the deformation modulus and Poisson’s ratio of the
rock mass below the shaft base. The proportion of the applied load transmitted to the
shaft base is
(7.21)
For side shear only shafts as shown in Figure 7.10(b), the boundary condition at the shaft
base is one of zero axial stress. For this case, the elastic displacement at the head of the
shaft can be obtained by
Fig. 7.10 Axially loaded drilled shafts

in rock (after Carter & Kulhawy,
1988).
Fig. 7.11 Comparison of analytical

solution with finite element solution
for predicting axial elastic
displacement (after Carter & Kulhawy,
1988).
(7.22)
The solution given by Equations (7.20) and (7.22) are in general agreement with the finite
element solutions by Pells and Turner (1979) and Rowe and Armitage (1987a) as
presented in last sections (Fig. 7.11).
Example 7.1
A drilled shaft of 3.0 meters long and 1.0 meter in diameter is to be installed in siltstone.
The rock properties are as follows:
RQD=70
Determine the settlement of the drilled shaft at a work load of 10.0 MN.
Solution:
For simplicity, the Young’s modulus of the drilled shaft is simply assumed to be Ep=30
GPa. The Poisson’s ratio of 0.25 is selected for both the drilled shaft and the rock.
Using Equation (4.24), the rock mass modulus:
αE=0.0231×70−1.32=0.297
Em=0.297×10.6=3.15Gpa
Using solutions based on finite element method
L/B=3.0/1.0=3.0
Ep/Em=30/3.15=9.52
If the drilled shaft is side shear resistance only (i.e., the shaft base cannot be cleaned),
from Figure 7.5, the axial deformation influence factor is I=0.462. Using Equation (7.15),
the settlement of the drilled shaft at the rock surface is
If the drilled shaft has both side shear and end bearing resistance, from Figure 7.8, the
axial deformation influence factor is I=0.417 for Eb/Em=1.0. Using Equation (7.15), the
settlement of the drilled shaft at the rock surface is
From Figure 7.9, it can be seen that about 15% of the load is transmitted to the shaft base.
Using analytical solutions of Carter and Kulhawy (1988)
ξ=Gb/Gm=1.0 for Eb/Em=1.0
If the drilled shaft is side shear resistance only (i.e., the shaft base cannot be cleaned), the
settlement of the drilled shaft at the rock surface can be calculated from Equation (7.22)
as
If the drilled shaft has both side shear and end bearing resistance, the settlement of the
drilled shaft at the rock surface can be calculated from Equation (7.20) as
The percentage of the load transmitted to the shaft base can be calculated from Equation
(7.21) as
The results from the solutions based on the finite element method are in good agreement
with those from the analytical solutions of Carter & Kulhawy (1988).
7.3.2 Nonlinear continuum approach
(a) Solutions based on finite element results

Rowe and Armitage (1987a) performed an elastic-plastic finite element analysis that
accounts for slip along the interface based on the technique developed by Rowe and Pells
(1980). Two layers of rock are considered in the analyses. The interface behavior is
established in terms of the Coulomb failure criterion. The roughness of the interface is
modeled implicitly through the use of an angle of interface dilatancy that produces
additional normal stress on the interface as the shaft deflects vertically due to the applied
load. The contribution of the interface dilatancy commences once slip occurs at the
interface. The results of this study are presented in three sets of design charts respectively
for Eb/Em=0.5, 1.0 and 2.0. Although the analysis is carried out considering the behavior
of a cohesive-frictional-dilative interface, the design charts are developed only for
nondilative-cohesive interfaces. The procedure for using the design charts is described in
Rowe and Armitage (1987b).
(b) Analytical solutions of Carter and Kulhawy (1988)

The case of slip along the entire length of the shaft has also been considered in detail by
Carter and Kulhawy (1988). For this case, the shear strength of the interface is given by
the Coulomb criterion:
(7.23)
where c is the interface cohesion; is the interface friction angle; and σr is the radial
stress acting on the interface.
As relative displacement (slip) occurs, the interface may dilate, and it is assumed that
the displacement components follow the dilation law:
(7.24)
where ∆u and ∆w are the relative shear and normal displacements of the shaft-rock
interface; and ψ is the angle of dilation defined by Davis (1968).
To determine the radial displacements at the interface, the procedure suggested by
Goodman (1980) and Kulhawy and Goodman (1987) is followed, in which conditions of
plane strain are assumed, as an approximation, independently in the rock mass and in the
slipping shaft. The rock mass is considered to be linear elastic, even after full slip has
taken place, and the shaft is considered to be an elastic column. These assumptions,
together with the dilatancy law, allow one to derive an expression for the variation of
vertical stress in the compressible shaft. The distribution of the shear stress acting on the
shaft can then be calculated from equilibrium conditions, and the vertical displacement
can be determined as function of depth z by treating the shaft as a simple elastic column.
The ‘full slip’ solution for the displacement of the shaft head is derived as
(7.25)
in which
F3=a1(λ1BC3−λ2BC4)−4a3
(7.26)
(7.27)
C3,4=D3,4/(D4−D3)
(7.28)
(7.29)
(7.30)
(7.31)
(7.32)
a1=(1+νm)ς+a2
(7.33)
(7.34)
(7.35)
All other parameters in Equations (7.25) to (7.35) are as defined before. The adequacy of
the closed-form expressions is demonstrated by comparing them with the finite element
solution of Rowe and Armitage (1987a, b). The overall agreement between the closed-
form solutions and the finite element results is good (Fig. 7.12).
It must be noted that the closed-form solutions of Carter and Kulhawy (1988) just
consider “no slip” (presented in Section 7.3.1) and “full slip” conditions. They cannot
predict the load-displacement response between the occurrence of first slip and full slip
of the shaft. However, the finite element results indicate that the progression of slip along
the shaft takes place over a relatively small interval of displacement. Therefore it seems
reasonable, at least for most practical cases, to ignore the small region of the curves
corresponding to the progressive slip and to assume that the load-displacement
relationship is bilinear, with the slope of the initial portion given by Equation (7.20) and
the slip portion by Equation (7.25) (Carter & Kulhawy, 1988).
Fig. 7.12 Axial displacement of a

drilled shaft in rock considering full
slip (after Carter & Kulhawy, 1988).
7.4 FINITE ELEMENT METHOD (FEM)
The finite element method is probably the most powerfiil and the most widely used
numerical method currently available to engineers. Suitable elements can be used to
simulate not only linearly elastic materials, but also nonlinear materials with different
failure criteria, including rock discontinuities and shaft-rock interfaces (see Sections 4.3.4
and 4.4.3 for discussion of joint elements). However, the finite element method is time
consuming and needs sophisticated soil or rock constitutive relations whose parameters
are often difficult if not impossible in design practice to obtain. Therefore, the finite
element method is, in general, used for analysis of important structures and for generation
of parametric solutions for the load-displacement relations of axially loaded drilled
shafts, such as the charts presented in Sections 7.3.1 and 7.3.2.
Typical of many geotechnical problems, the analysis of drilled shafts in rock involves
an unbounded domain. It is a common practice in finite element modeling of these
problems to truncate the finite element mesh at a distance deemed far enough so as not to
influence the near field solutions. These truncations are usually determined by trail and
error until an acceptable solution is obtained. Such a method places a heavy demand on
computer resources, both memory and time, as solutions for the far field which are of no
interest are generated as well. In the last decade or so, several methods have been
developed to model unbounded domains. Of these methods, the use of infinite elements
with finite elements appears to be the most popular. Leong and Randolph (1994)
successfully used finite elements and infinite elements in the modeling of axially loaded
shafts in rock.
7.5 DRILLED SHAFT GROUPS
Numerous methods exist for analyzing axially loaded pile groups in soil (Poulos, 2001),
some of which can be applied to drilled shaft groups in rock and are briefly described in
the following.
7.5.1 Settlement ratio method

In the settlement ratio method, the group settlement is related to the single-shaft
settlement as follows:
(7.36)
where wtG is the settlement of the shaft group; wtav is the settlement of a single shaft at the
average load of a shaft in the group; and Rw is the settlement ratio. wtav can be estimated
using the methods presented in the previous sections or from the results of load test on a
prototype drilled shaft.
Theoretical values of Rw for various pile groups in soil have been presented by Poulos
and Davis (1980) and Butterfield and Douglas (1981). A particularly useful
approximation for the settlement ratio has been derived by Fleming et al. (1992):
(7.37)
where n is the number of piles in the group; and e is an exponent depending on pile
spacing, pile proportions, relative pile stiffness and the variation of soil modulus with
depth. For typical pile proportions and pile spacings, Poulos (1989) suggested the
following approximate values: e≈0.5 for piles in clay, and e≈0.33 for piles in sand.
For drilled shafts in rock, the e values suggested by Poulos (1989) for soils may be
used for the very preliminary design. For the final design of major projects, it is desirable,
when feasible, to conduct axial load tests on groups of two or more drilled shafts in rock
in order to confirm the e values of Poulos (1989) or to derive new, site-specific values.
7.5.2 Equivalent pier method

The equivalent pier method, frequently used for pile groups in soils, treats the pile group
as an equivalent pier consisting of the piles and the soil between them (Poulos & Davis,
1980; Randolph, 1994). For closely spaced drilled shafts in rock, the shaft group may
also be analyzed using the equivalent pier method.
Consider the drilled shaft group as an equivalent pier (Fig. 7.13), the diameter of the
equivalent pier Beq can be taken as (Randolph, 1994).
Fig. 7.13 Equivalent pier method

treating drilled shaft group as a group
block.
(7.38)
where Ag is the plan area of the drilled shaft group as a block.

Deformation modulus of the equivalent pier Eeq is then calculated as
Eeq=Em+(Ep−Em)Apt/Ag
(7.39)
where Ep is the Young’s modulus of the drilled shafts; Em is the deformation modulus of
the rock mass; and Apt is the total cross-sectional area of the drilled shafts in the group.
The load-settlement response of the equivalent pier can be calculated using the
solutions as described in the previous sections for the response of a single drilled shaft.
Based on the equivalent pier method and the load-transfer (t-z curve) approach,
Castelli and Maugeri (2002) presented a simplified nonlinear analysis for settlement
prediction of pile groups in soil. To take into account the group action due to pile-soil-
pile interaction, load-transfer functions are modified to relate the behavior of a single pile
to that of a pile group. The bearing capacity of the equivalent pier can be evaluated using
the procedure in Section 6.4. The initial stiffness of the equivalent pier is estimated by
(7.40)
where Kgi is the initial stiffness of the equivalent pier and β is an empirical parameter. To
take into account the increase of pile group head settlements with respect to the case of a
single pile, the following expression is used
(7.41)
where wg is the average settlement of the equivalent pier and ε is an empirical parameter.
The empirical parameters β and ε can be derived on the basis of numerical analysis of
field tests. Castelli and Maugeri (2002) derived values of 0.30 and 0.15 respectively for β
and ε based on analysis of field test piles and pile groups in soils. For drilled shafts in
rock, similar values of β and ε can be obtained from field tests of shafts and shaft groups.
7.5.3 Finite element method (FEM)

The finite element method has been used to analyze axially loaded pile groups in soil by
simplifying the group to an equivalent plane strain or axisymmetric system. If necessary,
it can also be used to analyze drilled shaft groups in rock.
8
Lateral load capacity of drilled shafts in rock
8.1 INTRODUCTION
In the design of drilled shafts subjected to lateral forces, two criteria must be satisfied:
first, an adequate factor of safety against ultimate failure, second, an acceptable
deflection at working loads. This chapter discusses the prediction of ultimate load of
drilled shafts and drilled shaft groups. The calculation of lateral deflection will be
discussed in Chapter 9.
As the axial load capacity, the lateral load capacity of a drilled shaft in rock is
determined by the smaller of the two values: the structural strength of the shaft itself, and
the ability of the rock to support the loads transferred by the shaft.
8.2 CAPACITY OF DRILLED SHAFTS RELATED TO

REINFORCED CONCRETE
The structural capacity of a drilled shaft under lateral loading is controlled by the bending
capacity and the shear capacity. The bending capacity is usually checked by considering
the interaction between axial load and bending moment. Figure 8.1 shows the normalized
axial load-moment intersection diagrams for fy=10f′c and fy=15f′c, where fy is the yield
strength of the longitudinal reinforcing steel and f′c is the specified minimum concrete
strength. The factored axial load ΣγiQi is normalized by dividing by the factored nominal
axial capacity , where γi is the load factor for axial load i, Qi is the nominal value
of axial load i, and is the resistance factor for the nominal (computed) structural axial
load capacity Qu. The factored moment ΣγmMm is similarly normalized by dividing by the
factored nominal moment capacity , where γm is the load factor for moment m, Mm
is the nominal value of moment m, and is the resistance factor for the nominal
(computed) structural moment capacity Mu. The factored axial capacity is estimated from
Equation (6.1). Normalized axial load-moment interaction diagrams may be developed
for any fy/f′c ratios and cage diameters other than 0.6B.
With the axial load-moment interaction diagrams available, the structural capacity can
be checked as follows:
1. Estimate the combined axial load ΣγiQi.
2. Compute the factored nominal axial capacity from Equation (6.1).
Fig. 8.1 Normalized axial load-

moment interaction diagrams for
drilled shafts for (a)fy=10f′c and
(b)fy=15f′c.
Lateral load capacity of drilled shafts in rock 287
Table 8.1 Nominal structural moment capacity Mu

of drilled shaft.
Mu/f′cBAg
As/Ag fy−10f′c fy=15f′c
0.01 0.037 0.050
0.02 0.067 0.092
0.03 0.088 0.119
0.04 0.107 0.147
0.05 0.126 0.172
0.06 0.144 0.197
0.07 0.161 0.208
0.08 0.176 0.244
As is the cross-sectional area of the longitudinal reinforcing steel
Ag is the gross cross-sectional area of the shaft
fy is the yield strength of the longitudinal reinforcing steel
f′c is the specified minimum concrete strength
B is the diameter of the shaft
3. Estimate the factored (required) moment ΣγmMm.

4. Estimate the nominal structural moment capacity Mu of the drilled shaft. This may be
based on complete analysis, or it may be obtained from design aids such as Table 8.1.
5. Compute the factored nominal moment capacity where for
reinforced concrete.
6. Determine the ratios , and and with these values locate
an appropriate point on the axial load-moment interaction diagram. If the point falls
inside the area defined by the interaction curve, the shaft capacity is adequate. If this is
not the case, the shaft size should be increased and the analysis repeated until the shaft
capacity is adequate.
The factored nominal shear capacity of a drilled shaft without special shear reinforcement
can be calculated by (O’Neill & Reese, 1999):
(8.1)
where is the capacity reduction (resistance) factor for shear=0.85; Vu is the nominal
(computed) shear resistance; νc is the limiting concrete shear stress; and Av is the area of
the shaft cross section that is effective in resisting shear, which can be taken as
B(0.5B+0.5756rls) for a circular drilled shaft, where r1s is the radius of the ring formed by
the centroids of the longitudinal reinforcing steels. The limiting concrete shear stress νc
can be evaluated from:
(8.2a)
(8.2b)
where both f′c and νc are in kPa.

If the factored shear load is greater than the factored nominal shear resistance
determined above, two options are available. The first and simplest solution is to increase
the shaft diameter to increase the shear capacity. The second alternative is to provide
properly designed shear reinforcement (O’Neill & Reese, 1999).
O’Neill and Reese (1999) provide detailed discussion and examples on checking the
structural load capacity of drilled shafts.
8.3 CAPACITY OF DRILLED SHAFTS RELATED TO ROCK
8.3.1 Method of Carter and Kulhawy (1992)

Carter and Kulhawy (1992) presented a method to determine the lateral load capacity of
drilled shafts related to rock. When a lateral load is applied at the rock surface, the rock
mass immediately in front of the shaft will be subject to zero vertical stress, while
horizontal stress is applied by the leading face of the shaft. Ultimately, the horizontal
stress may reach the uniaxial compressive strength of the rock mass and, with further
increase in the lateral load, the horizontal stress may decrease as the rock mass softens
during postpeak deformation. Large lateral deformations may be required for the rock
mass at depth to exert a maximum reaction stress on the leading face of the shaft.
Therefore, Carter and Kulhawy (1992) assumed that the reaction stress at the rock mass
surface, in the limiting case of loading of the shaft, is zero or very nearly zero as a result
of the postpeak softening. Along the sides of the shaft, some shearing resistance may be
mobilized. The shearing resistance varies along the perimeter and the average can be
chosen as τmax/2, where τmax is likely to be approximately the same as the maximum unit
side resistance under axial compression. Therefore, at the rock surface, the ultimate force
per unit length resisting the lateral loading is Bτmax. At greater depth, Carter and Kulhawy
(1992) assume that the stress in front of the shaft increases from the initial in situ
horizontal stress to the limit stress, pL, reached during the expansion of a long cylindrical
cavity, i.e., a plane strain condition will apply. Behind the shaft, the horizontal stress will
decrease, and after tensile rupture of the bond between the concrete and the rock mass,
the horizontal stress will reduce to zero. At the sides of the shaft, some shearing
resistance may also be mobilized. Therefore, at depth, the ultimate force per unit length
resisting the lateral loading is B(pL+τmax). To determine the depth at which the limit stress
is mobilized, the result of Randolph and Houslby (1984) in a cohesive material is
adopted, i.e., the depth is about three times shaft diameter. Therefore, the distribution of
ultimate force per unit length resisting the shaft is as shown in Figure 8.2.
The ultimate lateral force that may be applied can be obtained from the horizontal
equilibrium as:
For L<3B
Fig. 8.2 Distribution of ultimate lateral

force per unit length (after Carter &
Kulhawy, 1992)
(8.3a)
For L>3B
(8.3b)
where τmax is the shearing resistance along the sides of the shaft, which is assumed to be
the same as the maximum side resistance under axial loading; pL is the limit stress
reached during the expansion of a long cylindrical cavity. Closed-form solutions have
been found for the limit stresses developed during the expansion of a long cylindrical
cavity in an elasto-plastic, cohesive-frictional, dilatant material (Carter et al., 1986). This
limit stress pL can be determined from the following parametric equation in the
nondimensional quantity ρ (Carter et al., 1986):
(8.4)
with
(8.5)
in which
(8.6)
(8.7)
(8.8)
(8.9)
(8.10)
(8.11)
(8.12)
(8.13)
(814)
and σhi is the initial in situ horizontal stress; Gm is the elastic shear modulus; νm is the
Poisson’s ratio; cm is the cohesion intercept; φm is the friction angle; and ψm is the dilation
angle, all of the rock mass. The rock mass is assumed to obey the Coulomb failure
criterion, and dilatancy accompanies yielding according to the following flow rule
(8.15)
in which dε1p and dε3p are the major and minor principal plastic strain increments,
respectively. For convenience, solutions for the limit pressures pL have been plotted in
Figure 8.3 for selected values of νm, φm, ψm. The central vertical axis on each plot
indicates the ratio of the plastic radius at the limit condition R to the cavity radius a.
These charts may be used by entering with a value of Gm/(σhi+cmcotφm) and working
clockwise around the figure, determining in turn values of R/a, then
ρL=(pL+cmcotφm)/(σR+ cmcotφm), and thus, determining the limit pressure pL.
8.3.2 Method of Zhang et al. (2000)

Zhang et al. (2000) presented an approximate method for calculating the ultimate lateral
resistance of drilled shafts in rock. As shown in Figure 8.4(a), the total reaction of the
rock mass consists of two parts: the side shear resistance and the front normal resistance.
So the ultimate resistance pult can be estimated by (Briaud & Smith, 1983; Carter &
Kulhawy, 1992):
(8.16)
Fig. 8.3 Limit solution for expansion

of cylindrical cavity (after Carter &
Kulhawy, 1992)
Fig. 8.4 (a) Components of rock mass

resistance; and (b) Calculation of
normal limit stress pL (after Zhang et
al., 2000).
Fig. 8.5 Distribution of ultimate lateral

resistance with depth (after Zhang et
al., 2000).
where B is the diameter of the shaft; τmax is the maximum shearing resistance along the
sides of the shaft; and pL is the normal limit resistance.
For simplicity, τmax is assumed to be the same as the maximum side resistance under
axial loading and can be determined using the methods presented in Chapter 6.
To determine the normal limit stress pL, the strength criterion for rock masses
developed by Hoek and Brown (1980, 1988) is used. Assuming that the minor principal
effective stress σ′3 is the effective overburden pressure γ′z and the limit normal stress pL
is the major principal effective stress σ′1 [see Fig. 8.4(b)], we have, from Equation (4.64),
the following
(8.17)
where γ′ is the effective unit weight of the rock mass; z is the depth from the rock mass
surface; and mb, s and a are rock mass parameters as described in Chapter 4.
With the distribution of pult along the depth determined, the ultimate lateral load that
may be applied can be approximated by (see Fig. 8.5)
(8.18)
Fig. 8.6 Comparison of estimated pult

and that from field shaft tests (after
Cho, 2002).
Cho (2002) used the method of Zhang et al. (2000) to estimate pult of field test shafts in
rock. The estimated values agree well with the field test data (see Fig. 8.6).
Example 8.1
A drilled shaft of diameter 1.0 m is to be installed 3.0 meters in siltstone. The rock
Unconfined compresive strength of intact rock, σc=15.0 Mpa
RMR=55
Determine the ultimate lateral load capacity of the shaft.
Solution:
From Table 4.5, mi=9 for siltstone.
Using Equation (4.68),
Using Equation (6.8) and choosing Ψ=1.0 (lower bound), the side shear resistance can
be obtained as
Assuming that the effective unit weight of the rock mass is 13.0 kN/m3, the limit
normal stress pL can be obtained from Equation (8.17) as follows
Using Equation (8.18), the ultimate lateral load capacity can be obtained as
It need be noted that the ultimate lateral load capacity obtained above does not
consider the moment equilibrium of the shaft. The structural strength of the shaft should
also be checked when using the ultimate lateral load capacity obtained above in design.
8.4 CAPACITY OF DRILLED SHAFT GROUPS
The ultimate lateral load capacity of a drilled shaft group can be calculated in a similar
way to calculating the axial load capacity of a drilled shaft group, i.e.
(8.19)
where HultG is the ultimate lateral load of a drilled shaft group; N is the number of drilled
shafts in the group; Hult is the ultimate lateral load of a single isolated drilled shaft; and α
is the group efficiency factor. Table 8.2 lists the values of α recommended by the
American Association of State Highway and Transportation Officials (AASHTO, 1989).
Equation (8.19) applies when the head boundary conditions of the single shaft and the
shaft group are the same. If the head boundary conditions of the single shaft and the shaft
group are different (e.g., the single shaft has a free-head boundary condition while the
shaft group has a flxed-head boundary condition because of the cap), a modification
factor should be added to Equation (8.19) to account for the difference in head boundary
conditions (Frechette et al., 2002):
HultG=αNHultR
(8.20)
where R is the modification factor to account for the difference in head boundary
conditions. For the case of a single shaft at a free-head boundary condition and a shaft
group at a fixed-head boundary condition, Frechette et al. (2001) recommended a R value
of 2.2 based on five case studies while Matlock and Foo (1976) recommended a R value
of 2.0 based on a single case study (Frechette et al., 2002).
If the drilled shafts are closely spaced, the drilled shaft group can be represented by a
group block and its ultimate load can be calculated using the methods described in
Section 8.3 by treating the group block as a big single shaft.
8.5 DISCONTINUUM METHOD
In Sections 8.3 and 8.4, the rock mass is treated as a continuum. Since most rocks contain
discontinuities, drilled shafts may fail due to the sliding of the rock blocks or wedges
along discontinuities (see Fig. 8.7). In such cases, the lateral resistance is only provided
by the shear resistance along the discontinuities and the weight of wedge bounded by the
shaft and the discontinuities. Obviously, the rock mass need be treated as a discontinuous
medium in order to obtain the lateral resistance provided by the wedges.
Table 8.2 Group efficiency factor a recommended
by AASHTO (1989).
Center-to-Center Shaft Spacing in Group Efficiency Factor
Direction of Loading α
3B 0.25
4B 0.40
6B 0.70
8B 1.00
Figure 8.7 Sliding of rock blocks due

to laterally loaded shaft.
To (1999) developed a discontinuum method for determining the lateral load capacity of
drilled shafts in a jointed rock mass containing two or three discontinuity sets. The
method consists of two parts: a kinematic analysis and a kinetic analysis. In the kinematic
analysis, Goodman’s block theory (Goodman & Shi, 1985) is extended to analyze the
movability of a combination of blocks laterally loaded by a drilled shaft. Based on the
extended theory, a 2-dimensional (2D) graphical method was developed to select the
possible combinations of movable blocks. This 2D graphical method can be easily
implemented with CAD programs such as AutoCAD or with spreadsheet programs such
as Excel.
In the kinetic analysis, the stability of each kinematically selected movable
combination of blocks or wedges is analyzed with the limit equilibrium approach. This
analysis, similar to slope stability analysis, considers the axial and lateral forces exerted
by the drilled shaft in addition to the weight of the wedge and the shearing resistance
along the discontinuities. From the stability analysis, simple analytical relations were
developed to solve for the lateral load capacity of the drilled shaft.
The lateral load capacity can also be obtained by analyzing the load-displacement
response of a drilled shaft using the discrete element method (DEM) as described in
Section 9.5.
9
Lateral deformation of drilled shafts in rock
9.1 INTRODUCTION
For drilled shafts in rock to resist lateral loads, the design criterion in the majority of
cases is not the ultimate lateral capacity of the shafts, but the maximum deflection of the
shafts. Predicting the deformation of laterally loaded drilled shafts is, therefore, the most
important aspect in designing drilled shafts to withstand lateral loads.
To date, it has been customary practice to adopt the techniques developed for laterally
loaded piles in soil (Poulos, 1971a, b, 1972; Banerjee & Davies, 1978; Randolph, 1981)
to solve the problem of drilled shafts in rock under lateral loading (Amir, 1986; Gabr,
1993; Wyllie, 1999). However, the solutions for laterally loaded piles in soil do not cover
all cases for laterally loaded drilled shafts in rock in practice (Carter & Kulhawy, 1992).
Carter & Kulhawy (1992), therefore, developed a method for predicting the deformation
of laterally loaded drilled shafts in rock. This method treats the rock mass as an elastic
continuum and has been found to give reasonable results of predicted deflections only at
low load levels (20–30% capacity). At higher load levels, the predicted displacements are
too small (DiGioia & Rojas-Gonzalez, 1993). Reese (1997) developed a p-y curve
method for analyzing drilled shafts in rock under lateral loading. The major advantage of
the p-y curve approach lies in its ability to simulate the nonlinearity and nonhomogeneity
of the rock mass surrounding the drilled shaft. However, since it represents the rock mass
as a series of springs acting along the length of the shaft, the p-y curve approach ignores
the interaction between different parts of the rock mass. Also, the p-y curve approach
uses empirically derived spring constants that are not measurable material properties.
Advances in computer technology have made it possible to analyze laterally loaded piles
using three-dimensional (3D) finite element (FE) models. p-y curves (Hoit et al., 1997) or
sophisticated constitutive relations (Wakai et al., 1999) are usually used to represent the
soil or rock behavior in the 3D FE analyses. However, p-y curves have the limitations as
described above. As for sophisticated soil or rock constitutive relations, it is often
difficult if not impossible in design practice to obtain the parameters in the constitutive
relations.
Zhang et al. (2000) developed a nonlinear continuum method for analyzing laterally
loaded drilled shafts in rock. The method can consider drilled shafts in a continuum
consisting of a soil layer overlying a rock mass layer. The deformation modulus of the
soil is assumed to vary linearly with depth while the deformation modulus of the rock
mass is assumed to vary linearly with depth and then stay constant below the shaft tip.
The effect of soil and/or rock mass yielding on the behavior of shafts is considered by
assuming that the soil and/or rock mass behaves linearly elastically at small strain levels
and yields when the soil and/or rock mass reaction force exceeds the ultimate resistance.
9.2 SUBGRADE-REACTION (p-y CURVE) APPROACH
Treating the rock as a series of springs along the length of the shaft (see Fig. 9.1), the
behavior of the shaft under lateral load can be obtained by solving the following
differential equation (Reese, 1997)
where Q is the axial load on the shaft; y is the lateral deflection of the shaft at a point z
along the length of the shaft; p is the lateral reaction of the rock; EpIp is the flexural
rigidity of the shaft; and W is the distributed horizontal load along the length of the shaft.
Equation (9.1) is the standard beam-column equation where the values of EpIp may
change along the length of the shaft and may also be a function of the bending moment.
The equation (a) allows a distributed load to be placed along the upper portion of a shaft;
(b) can be used to investigate the axial load at which a shaft will buckle; and (c) can deal
with a layered profile of soil or rock (Reese, 1997).
Computer programs, such as COM624P and LPILE, are available to solve equation
(9.1) efficiently. COM624P (version 2.0 and higher) and LPILEPLUS can also consider the
variation of EpIp with the bending moment (see O’Neill & Reese, 1999 for the detailed
procedure). To solve Equation (9.1), boundary conditions at the top and bottom of the
shaft also need be considered. For example, the applied shear and moment at the shaft
head can be specified, and the shear and moment at the base of the shaft can be taken to
be zero if the shaft is long. For short shafts, a base boundary condition can be specified
that allows for the imposition of a shear reaction on the base as a function of lateral base
deflection. Full or partial head restraint can also be specified. Other formula that are used
in the analysis are
(9.2)
(9.3)
(9.4)
where V, M and S are respectively the transverse shear, bending moment and deflection
slope of the drilled shaft.
The major difference between various methods lies in the determination of the
variation of p with y or the p-y curve, which are described below.
Lateral deformation of drilled shafts in rock 301
Fig. 9.1 Subgrade-reaction (p-y curve)

model of laterally loaded drilled shafts.
9.2.1 Linear analysis

For linear analysis, the relationship between rock reaction p and shaft deflection y at any
point along the shaft is assumed to be linear, i.e.,
(9.5)
where kh is the coefficient of subgrade reaction, in the unit of force/length3; and B is the
width or diameter of the shaft. Substituting Equation (9.5) into Equation (9.1) and
neglecting the influence of Q and W, the governing equation for the deflection of a
laterally loaded shaft with constant EpIp can be simplified as
(9.6)
Solutions to the above equation may be obtained analytically as well as numerically with
a computer program.
The analysis of the load-displacement behavior of a drilled shaft also requires
knowledge of the variation of kh along the shaft. A number of distributions of kh along the
depth have been employed by different investigators, which can be described by the
following general expression proposed by Bowles (1996):
kh=Ah+Bhzn
(9.7)
where Ah, Bh and n are empirical constants which can be determined for a particular site
by working backward from the results of lateral shaft load tests.
If the rock is considered homogeneous with a constant kh down the length of the
drilled shaft, the deflection u (both y and u are used to denote lateral deflection in this
book) and rotation θ at the ground level due to applied load H and moment M can be
calculated by
(9.8a)
(9.8b)
where Lc is the critical length given by
(9.9)
It should be noted that Equation (9.8) is applicable only to flexible shafts, i.e., shafts
longer than their critical length defined by Equation (9.9). For non-flexible shafts,
solutions in closed-form expressions or in the form of charts are also available
(Tomlinson, 1977; Reese & Van Impe, 2001).
9.2.2 Nonlinear analysis

In general, the relationship between rock reaction p and shaft deflection y at any point
along a shaft is nonlinear. Kubo (1965) used the following nonlinear relationship for soil
between reaction p, deflection y, and depth z:
p=kzmyn
(9.10)
where k, m, and n are experimentally determined coefficients. Equation (9.10) can also be
used for rock if the corresponding coefficients k, m, and n can be determined.
Since Matlock (1970) developed a method for deriving the variation of p with y, or the
p-y curves, for soft clay, based on field test results, a number of methods for deriving p-y
curves for different soils have been developed. Some of them are listed below (for details,
the reader can refer to the listed references):
1.API RP2A (1982) or Reese et al. (1974) method for sand.
2. Bogard and Matlock (1980) method for sand.

3. API RP2A (1991) or O’Neill and Murchison (1983) method for sand.
4. API RP2A (1982) or Matlock (1970) method for soft clay.
5. API RP2A (1982) or Reese et al. (1975) method for stiff clay.
6. Integrated method for clay by Gazioglu and O’Neill (1984).
7. Pressuremeter methods for all soils (Robertson et al., 1982, 1986; Briaud & Smith,
1983; Briaud, 1986).
The method, developed by Reese (1997), specifically for calculating the p-y curves for
rock is described in the following section.
9.2.5 p-y curves for rock

Reese (1997) presented a p-y curve method for analyzing laterally loaded drilled shafts in
rock. The concepts and procedures for constructing the p-y curves for rock are as follows
(Reese, 1997):
(1) The secondary structure of rock, related to joints, cracks, inclusions, fractures, and
any other zones of weakness, can strongly influence the behavior of the rock and thus
need be taken into account when applying the method described in this section.
(2) The p-y curves for rock and the bending stiffness E0Ip for the shaft must both reflect
nonlinear behavior in order to predict loadings at failure.
(3) The initial slope Kmi of the p-y curves must be predicted because small lateral
deflections of shafts in rock can result in resistances of large magnitudes. For a given
value of compressive strength, Kmi is assumed to increase with depth below the ground
surface.
(4) The modulus of the rock Em, for correlation with Kmi, may be taken from the initial
slope of a pressuremeter curve. Alternatively, the correlations presented in Chapter 4
can be used to determine Em.
(5) The ultimate resistance pult for the p-y curves will rarely, if ever, be developed in
practice, but the prediction of pult is necessary in order to reflect nonlinear behavior.
(6) The component of the strength of rock from unit weight is considered to be small in
comparison to that from compressive strength, and therefore the weight of rock is
ignored.
(7) The compressive strength σc of the intact rock for computing pult may be obtained
from tests of intact specimens.
(8) The assumption is made that fracturing will occur at the surface of the rock under
small deflections; therefore, the compressive strength of intact rock specimens is
reduced by multiplication by αm to account for fracturing. The value of αm is assumed
to be 1/3 for RQD of 100 and to increase linearly to unity at RQD of zero. If RQD is
zero, the compressive strength may be obtained directly from a pressuremeter curve.
(a) Calculation of ultimate resistance pult of rock

The following expressions are used for calculating the ultimate resistance pult of rock
(9.11a)
(9.11b)
where B is the diameter of the shaft; zm is the depth below the rock surface; σc is the
unconfined compressive strength of the intact rock; and αm is the strength reduction factor
considering that fracturing will occur at the surface of the rock under small deflections
and thus reducing the resistance of the rock.
(b) Calculation of the slope of initial portion of p-y curves

The slope of the initial portion of p-y curves, kmi, is estimated by
Kmi=kmiEm
(9.12)
where Em is the modulus of the rock (mass); and kmi is a dimensionless constant which
can be determined by
(9.13a)
kmi=500 zm≥3B
(9.13b)
Equation (9.13) is developed from experimental data and reflect the assumption that the
presence of the rock surface has a similar effect on kmi, as was shown for the ultimate
resistance pult.
(c) Calculation of p-y curves

Referring to Figure 9.2, the p-y curve consists of three portions. The initial and the third
portions are straight-lines and the second portion is a curve. The three portions can be
expressed by
First Portion: p=Kmiy; y≤yA
(9.14a)
(9.14b)
Third Portion p=pult

(9.14c)
in which ym=kmB, where km is a constant, ranging from 0.0005 to 0.00005, that serves to
establish overall stiffness of curves. The value of yA is found by solving the intersection
of Equations (9.14a) and (9.14b), and is shown by
(9.15)
Fig. 9.2 Sketch of p-y curve for rock

(after Reese, 1997).
(d) Comments
The equations described above for constructing the p-y curves for rock are based on
limited data and should be used with caution. An adequate factor of safety should be
employed in all cases; preferably, field tests should be undertaken on full-sized shafts
with appropriate instrumentation. If the rock contains joints that are filled with weak soil,
the selection of strength and stiffness must be site-specific and will require a
comprehensive geotechnical investigation. In those cases, the application of the method
presented in this section should proceed with even more caution than normal (Reese,
1997).
Cho et al. (2001) conducted lateral load tests on two drilled shafts embedded in
weathered Piedmont rock. These shafts were instrumented with inclinometers and strain
gauges. The field data obtained from the instrumented shafts were used to backcalculate
the p-y curves. A comparison of the back-calculated p-y curves with the p-y curves
predicted using the method of Reese (1997) shows that the method of Reese (1997)
significantly overestimates the resistance of the weathered rock.
9.3 CONTINUUM APPROACH
The continuum approach assumes the soil and rock to be a continuum. Numerical
solutions were developed by assuming that the soil and rock are ideally elastic, first with
the boundary element method (Poulos, 1971a, b, 1972; Banerjee & Davies, 1978) and
second with the finite element method (Randolph, 1981). Most of these elastic solutions
were presented in the form of charts. Randolph (1981) published approximate but
convenient closed-form expressions for the response of flexible piles to lateral loading.
Considering the fact that the closed-form expressions of Randolph (1981) for the lateral
response of flexible piles in soils may not cover the ranges of material and geometric
parameters encountered in drilled shafts in rock, Carter and Kulhawy (1992) expanded
the solutions by Randolph (1981). The solutions of Carter and Kulhawy (1992) give a
reasonable agreement between measured and predicted displacements for drilled shafts in
rock at low load levels (20–30% capacity). At higher load levels, however, the predicted
displacements are too small (DiGioia & Rojas-Gonzalez, 1993). Zhang et al. (2000)
developed a nonlinear continuum approach for the analysis of laterally loaded drilled
shafts in rock. The approach can consider the effect of soil and/or rock mass yielding on
the behavior of shafts.
9.3.1 Linear continuum approach
(a) Approach of Poulos (1971a, b, 1972) and Poulos and Davis (1980)
By modeling the soil as an elastic continuum and idealizing the pile as an infinitely thin
strip of the same width and bending rigidity as the prototype pile, Poulos (1971a, b, 1972)
and Poulos and Davis (1980) obtained the solutions for laterally loaded piles using the
boundary element method. The solutions are presented in the form of charts and can be
used to predict the deflection of drilled shafts in rock.
For a free head drilled shaft, the lateral deflection u and rotation θ under lateral force
H and overturning moment M at ground surface are given by
(9.16a)
(9.16b)
where L is the length of the shaft; EmL is the deformation modulus of the rock mass at the
level of shaft tip; and IuH, IuM, IθH and IθM (note that IuM=IθH) are deflection and rotation
influence factors which are a function of the drilled shaft flexibility factor KR and the
rock mass non-homogeneity η:
(9.17)
(9.18)
where Em0 is the deformation modulus of the rock mass at the ground surface. A
homogeneous rock mass is represented by η=1, whereas η=0 represents a rock mass with
zero modulus at the surface. The deflection and rotation influence factors are plotted in
Figures 9.3 to 9.5 for values of η of 0 and 1. If the shaft is partially embedded, the
deflection of the free-standing portion due to shaft rotation and bending can be added to
the groundline deflection to obtain the deflection at the shaft head.
If the drilled shaft is fixed-headed, the horizontal deflection can be obtained by putting
θ =0 in Equation (9.16b) and substituting for the obtained moment in Equation (9.16a), as
(9.19)
Fig. 9.3 Deflection influence factor IuH

(after Poulos & Davis, 1980)
Fig. 9.4 Deflection and rotation

influence factors IuM and IθH (after
Poulos & Davis, 1980)
Fig. 9.5 Rotation influence factor IθM

(after Poulos & Davis, 1980).
For a single raking shaft, Poulos and Madhav (1971) have shown that the force acting on
the shaft head may be resolved into axial and normal components and the shaft then
treated as a vertical shaft subjected to these forces and the applied moment.
(b) Approach of Randolph (1981) and Carter and Kulhawy (1992)

Randolph (1981) conducted a parametric study of the response of laterally loaded piles
embedded in an elastic soil continuum. The study was conducted using the finite element
method and the results were fitted with closed-form expressions from which the lateral
response of piles may be readily calculated. Considering the fact that the closedform
expressions for the lateral response of flexible piles in soils may not cover the ranges of
material and geometric parameters encountered in drilled shafts in rock, Carter and
Kulhawy (1992) expanded the solutions by Randolph (1981). The expressions were
derived from the results of finite element studies of the behavior of laterally loaded
drilled shafts in rock. For a drilled shaft wholly embedded in rock [Fig. 9.6(a)], the shaft
response can be calculated in the following way (Carter & Kulhawy, 1992):
(1) The shaft is considered flexible when
(9.20)
where Ee is the effective Young’s modulus of the shaft
(9.21)
in which B and EpIp are respectively the diameter and flexural rigidity of the shaft; and
G* is the equivalent shear modulus of the rock mass
(9.22)
in which Gm and νm are respectively the shear modulus and Poisson’s ratio of the rock
mass.
The shaft response can then be obtained by the closed-form expressions suggested by
Randolph (1981), i.e.,
(9.23a)
(9.23b)
Fig. 9.6 (a) Drilled shaft wholly

embedded in rock; and (b) Drilled
shaft embedded in soil and rock.
(2) The shaft is considered rigid when
(9.24)
The shaft response can then be obtained by the following closed-form expressions
(9.25a)
(9.25b)
(3) The shaft can be described as having intermediate stiffness whenever the slenderness
ratio is bounded approximately as follows
(9.26)
The finite element results show that the displacements for an intermediate
case exceed the maximum of the predictions for corresponding rigid and
flexible shafts by no more than about 25%, and often by much less. For
simplicity, it is suggested that the shaft displacement in the intermediate
case be taken as 1.25 times the maximum of either: (a) The predicted
response of a rigid shaft with the same slenderness ratio L/B as the actual
shaft; or (b) the predicted response of a flexible shaft with the same
modulus ratio (Ee/G*) as the actual shaft. Values calculated in this way
should, in most cases, be slightly larger than those given by the more
rigorous finite element analysis for a shaft of intennediate stiffness.
If there exists a layer of soil overlying rock as shown in Figure 9.6(b), Carter and
Kulhawy (1992) assume that the complete distribution of soil reaction on the shaft is
known and that the socket provides the majority of resistance to the lateral load or
moment. The groundline horizontal displacement u and rotation θ can then be determined
after structural decomposition of the shaft and its loading, as shown in Figure 9.7. To
determine the distribution of the soil reaction, they simply assume that the limiting
condition is reached at all points along the shaft, from the ground surface to the interface
with the underlying rock mass, and then use the reaction distribution suggested by Broms
(1964a, b).
For shafts through cohesive soils (Fig. 9.8), the lateral displacement uAO and rotation
θAO of point A relative to point O are given by
(9.27a)
Fig. 9.7 Consideration of soil reaction:

(a) Loading and displaced shape; and
(b) Decomposition of loading (after
Carter & Kulhawy, 1992).
(9.27b)
where Ls is the thickness of the soil layer; and su is the undrained shear strength of the
soil. The shear force Ho and bending moment Mo at point O are determined by
HO=H−9su(Ls−1.5B)B
(9.28a)
2
MO=M−4.5su(Ls−1.5B) B+HLs
(9.28b)
The contribution to the groundline displacement from the loading transmitted to the rock
mass can then be computed by analyzing a fully rock-socketed shaft of embedded length
L, subject to horizontal force HO and moment MO applied at the level of the rock mass.
For shafts through cohesionless soils (Fig. 9.9), the lateral displacement uAO and
rotation θAO of point A relative to point O are given by
(9.29a)
(9.29b)
where γ′ is the effective unit weight of the soil; and Kp is the Rankine passive earth
pressure coefficient. The shear force HO and bending moment MO at point O are
determined by
(9.30a)
Fig. 9.8 Idealized loading of socketed

shaft through cohesive soil (after
Fig. 9.9 Idealized loading of socketed

shaft through cohesionless soil (after
(9.30b)
Example 9.1
A drilled shaft of diameter 1.0 m is to be installed 3.0 meters in siltstone. The rock
Unconfined compressive strength of intact rock, σc=15.0 Mpa
Deformation modulus of intact rock Er=10.6 GPa
RQD=70
Determine the lateral displacement and rotation of the drilled shaft at the groundline
by a horizontal force of 2.6 MN at 2.5 m above the groundline.
Solution:
For simplicity, the Young’s modulus of the drilled shaft is simply assumed to be
Ep=30 GPa. A Poisson’s ratio of 0.25 is selected for both the drilled shaft and the rock.
The flexural rigidity of the shaft is
Using Equation (4.24), the deformation modulus of the rock mass is

αE=0.0231×70−1.32=0.297
Em=0.297×10.6=3.15 Gpa
and the shear modulus of the rock mass is

Gm=3.15/(1+0.25)=1.26 Gpa
Using Equation (9.22), the equivalent shear modulus of the rock mass is
G*=1.26×(1+3×0.25/4)=1.50 Gpa
Since
the shaft is considered flexible and the lateral displacement and rotation of the drilled
shaft at the groundline can be obtained from Equation (9.23) as
9.3.2 Nonlinear continuum approach

Poulos and Davis (1980) presented an approximate nonlinear approach for calculating the
deflection of laterally loaded piles in soil. This approach uses the elastic solutions
presented in the last section, but introduces yield factors. The yield factors are a function
of relative flexibility and load level and allow for the increased deflection and rotation of
a pile due to the onset of local yielding of the soil adjacent to the pile. This approach can
also be used to calculate the nonlinear deflection of drilled shafts in rock.
For a drilled shaft subjected to a lateral load H at an eccentricity of e above the

groundline, the groundline deflection u and rotation θ can be expressed as follows:
(1) Uniform modulus with depth, i.e., η=1.0
(9.31a)
(9.31b)
(2) Linearly increasing modulus with depth, i.e., η=0
(9.32a)
(9.32b)
where uelastic and θelastic are respectively deflection and rotation from elastic solutions as
described in the previous section; and Fu, Fθ, F′u, and Fθ are yield deflection and rotation
factors which can be found from Poulos and Davis (1980). The yield factors are functions
of a dimensionless load level H/Hu, where Hu is the ultimate lateral load capacity of the
equivalent rigid shaft and can be estimated using the methods presented in Chapter 8.
Zhang et al. (2000) developed a nonlinear continuum approach for the analysis of
laterally loaded drilled shafts in rock. This approach adopts and extends the basic idea of
Sun’s (1994) work on laterally loaded piles in soil. Sun’s model treats soil as a
homogeneous elastic continuum with a constant Young’s modulus, which may apply to
stiff clay, and it does not consider yielding of the soil. In the nonlinear approach
developed by Zhang et al. (2000), drilled shafts in a soil and rock mass continuum (see
Fig. 9.10) are considered, and the effect of soil and/or rock mass yielding on the behavior
of shafts is included. For simplicity, the shaft is assumed to be elastic, while the soil/rock
mass can be either elastic or elasto-plastic. It is, nevertheless, possible to also check
whether the shaft concrete will yield or not using standard concrete design methods, as
will be briefly mentioned later.
(a) Method of analysis—elastic behavior
Governing equations of shaft and soil/rock mass system

Consider a drilled shaft of length L, radius R and flexural rigidity EpIp, embedded within
a
Fig. 9.10 (a) Shaft and soil/rock mass

system; (b) Coordinate system and
displacement components; and (c)
Shear force V(z) and moment M(z)
acting on shaft at z (after Zhang et al.,
2000).
soil/rock mass system (Fig. 9.10). The deformation modulus of the soil varies linearly
from Es1 at the ground surface to Es2 at the soil/rock mass interface. The deformation
modulus of the rock mass varies linearly from Em1 at the soil/rock mass interface to Em2 at
the shaft tip and stays constant below the shaft tip. For convenience of presentation, non-
uniformity indices defined by
(9.33)
(9.34)
are introduced. The increase of the deformation moduli of the soil and the rock mass with
depth, z, can then be expressed, respectively, by
(9.35a)
(9.35b)
Em=Em2 (z>Ls+L)
(9.35c)
By adopting the basic idea of Sun (1994), the displacements usm, νsm and usm of the soil
and/or rock mass can be approximated by separable functions of the cylindrical
coordinates r, θ and z as
(9.36a)
(9.36b)
wsm(r,θ,z)=0
(9.36c)
where u(z) is the displacement of the shaft as a function of depth; and is a

dimensionless function representing the variation of displacements of the soil and/or rock
mass in the r-direction.
For the displacements of Equation (9.36), the governing equations for the shaft can be
obtained as
(9.37a)
(9.37b)
with boundary conditions
(9.38a)
(9.38b)
(9.38c)
us−um=0 (z=Ls)
(9.38d)
(9.38e)
(9.38f)
(9.38g)
(9.38h)
(9.38i)
where us and um are the displacement components u of the shaft in the soil and in the rock
mass, respectively; and ts, ks and tm, km are parameters that can be expressed as
(9.39a)
(9.39b)
(9.39c)
(9.39d)
where m1 and m2 are parameters describing the behavior of the elastic foundations, which
can be obtained by
(9.40a)
(9.40b)
Function can be obtained by solving the following equation

(9.41)
where γ is a nondimensional parameter that can be expressed as
(9.42)
The solution to Equation (9.41) that satisfies the unit condition at and
the finite condition at can be obtained and the parameters m1
and m2 can then be expressed as (Sun, 1994)
(9.43a)
(9.43b)
where K0( ) is the modified Bessel function of the second kind of zero order; and K1( ) is
the modified Bessel function of the second kind of first-order.
The shear force V(z) acting on the shaft (see Fig. 9.10) can be obtained by
(9.44a)
(9.44b)
and the bending moment M(z) acting on the shaft (see Fig. 9.10) can be obtained by
(9.45a)
(9.45b)
The governing differential equations and the shear force V(z) and bending moment M(z)
are solved using the classical finite difference method as described below. At this point it
is also possible to check if the shaft concrete yields (recall that the basic assumption is
non-yielding concrete). This can be done using the calculated shear force and moment
together with the axial force on the shaft and using standard concrete design methods.
Finite difference model

The classical finite difference method (Desai & Christian, 1977) is employed to solve the
governing differential Equation (9.37). By dividing the shaft in the soil into Ns equal
segments (see Fig. 9.11) and using the central difference operator, for an interior node i
(i= 0, 1, 2,…, Ns), the following equation is obtained:
Fig. 9.11 Dividing shaft into segments

for finite-difference analysis, and
estimating reaction force p of soil and
rock from shear force V (after Zhang et
al., 2000).
(9.46)
where
(9.47)
in which hs=Ls/Ns.
Similarly, by dividing the shaft in the rock mass into Nm equal segments (see Fig.
9.11), the following equation is obtained for an interior nodey j(j=0, 1, 2,…, Nm):
(9.48)
where
(9.49)
in which hm=L/Nm.
Equations (9.46) and (9.48) can be written recursively for each point i=0, 1, 2,…, Ns
and j=0, 1, 2,…, Nm(see Fig. 9.11), resulting in a set of simultaneous equations in u. To
solve the set of equations the boundary conditions must be introduced. By incorporating
the boundaiy conditions expressed by Equation (9.38), the following finite difference
equations can be obtained:
at z=0
(9.50a)
(9.50b)
−us(−1)−us(1)=0 (fixed-head)
(9.50c)
at z=Ls
(9.50d)
(9.50e)
(9.50f)
(9.50g)
at z=Ls+L
(9.50h)
(9.50i)
The set of equations [Equations (9.46) and (9.48)] is modified by introducing the
boundary conditions given in Equation (9.50). The resulting equations are solved
simultaneously for u by using the Gaussian elimination procedure.
After the shaft displacement u is obtained, the shear force V acting on the shaft can be
obtained from Equation (9.44) as:
(9.51a)
(9.51b)
The bending moment M acting on the shaft can be obtained from Equation (9.45) as:
(9.52a)
(9.52b)
With the shear force V(z) obtained from Equation (9.51), the lateral reaction force p(z)
(F/L) of the soil and rock mass acting on the shaft can be estimated by (see Fig. 9.11)
(9.53a)
(9.53b)
Iteration procedure
To solve for u, parameters ts, ks and tm, km should be known [see Equations (9.46) to
(9.50)]. As can be seen from Equations (9.39) and (9.43), the parameter γ is needed to get
ts, ks and tm, km. Note that γ defined by Equation (9.42) depends on u. Since we do not
know the value of γ a priori, an iterative procedure is required to obtain it (Sun, 1994).
The procedure consists of the following steps:
1. Assume γ=1.0
2. Calculate m1 and m2 from Equation (9.43)
3. Calculate ts, ks and tm, km from Equation (9.39)
4.Calculate the pile displacement u(z) along the shaft by solving Equations (9.46), (9.48)
and (9.50)
5. Calculate the new value of γ using Equation (9.42)
6. Use the new value of γ and repeat steps 2–5. The iteration is continued until the ith and
(i+1)th γ meet following criterion:
(9.54)
where ε is a prescribed convergence tolerance, say, 0.0001. After γ is determined,

the displacement of the shaft can be obtained.
7. Calculate the shear force and bending moment distribution along the shaft from
Equations (9.51) and (9.52).
8. Calculate the lateral reaction force p(z) of the soil and rock mass acting on the shaft
from Equation (9.53).
(b) Method of analysis—including yielding of soil and rock mass
Consideration of yielding of soil and rock mass

For a laterally loaded shaft, the soil or rock mass near the top of the shaft may yield if the
loads are large enough and, consequently, increased displacements will occur. Hence it is
important to consider the effect of yielding of the soil or rock mass on the shaft behavior.
A simple method is proposed to consider local yielding of the soil and rock mass by
assuming that the soil and rock mass are elastic-perfectly plastic. The method consists of
the following steps (Fig. 9.12):
1. For the applied lateral load H and moment M the shaft is analyzed by assuming the soil
and rock mass are elastic and the lateral reaction force p of the soil and rock mass
along the shaft is determined as described in the elastic analysis.
2. Compare the computed lateral reaction force p with the ultimate resistance pult (which
will be discussed in detail in next section) and, if p>pult, determine the yield depth, zy,
in the soil and/or rock mass.
Fig. 9.12 Consideration of yielding of

soil and/or rock by decomposition of
loading (after Zhang et al., 2000).
3. Consider the portion of the shaft in the unyielded ground (soil and/or rock mass) (zy
≤z≤Ls+L) as a new shaft and analyze it by ignoring the effect of the soil and/or rock
mass above the level z=zy. The lateral load and moment at the new shaft head are
(9.55a)
(9.55b)
4. Repeat steps 2–3. The iteration is continued until no further yielding of soil or rock
mass occurs.
5. Obtain the final results by considering the two parts of the shaft separately. The part in
the yielded soil and/or rock mass is analyzed as a beam with the distributed load pult
acting on it. The part in the unyielded soil and/or rock mass is analyzed as a shaft with
the soil and/or rock mass behaving elastically.
A computer program has been written to execute the above iteration procedure including
the process of elastic analysis.
Determination of ultimate resistance pult of soil and rock mass

To consider the yielding of soil and rock mass, the ultimate resistance pult of the soil and
rock mass need be determined. For clays, it is usual to adopt a total stress approach and
consider the ultimate soil resistance under undrained loading conditions. The simplest
approach is to express pult as follows:
pult=NpsuB
(9.56)
where B is the diameter of the shaft; su is the undrained shear strength of soil; and Np is
the bearing capacity factor. A number of expressions for estimating Np are available in
the literature (Hansen, 1961; Broms, 1964a; Matlock, 1970; Reese & Welch, 1975;
Stevens & Audibert, 1979; Randolph & Houlsby, 1984). Zhang et al. (2000)
recommended the following expression for Np, which was proposed by Matlock (1970)
and Reese and Welch (1975) and are most widely employed in engineering practice:
(9.57)
where γ′ is the average effective unit weight of soil above depth z; and J is a coefficient
ranging from 0.25 to 0.5.
For sand, several methods are available in the literature for estimating pult (Broms,
1964b; Reese et al., 1974; Borgard & Matlock, 1980; Fleming et al., 1992). These
methods often produce significantly different values of pult (Zhang et al., 2002). By
analyzing the lateral soil resistance distribution along the width of piles and based on the
test results of model rigid piles in sand collected from the published literature, Zhang et
al. (2002) developed the following expression for calculating pult
(9.58)
where B is the diameter of the shaft; pmax is the maximum normal resistance against the
shaft; τmax is the maximum shear resistance against the shaft; and η and β are the shape
factors to account for the non-uniform distribution of the normal resistance and the shear
resistance along the width of piles. According to Briaud et al. (1983), η and β can be
respectively taken as 0.8 and 1.0 for round piles. pmax and τmax are calculated by (Zhang et
al., 2002)
(9.59)
(9.60)
where γ′ is the average effective unit weight of soil above depth
is the Rankine passive earth pressure coefficient, in which

is the effective internal friction angle; K is the coefficient of lateral earth pressure (ratio
of horizontal to vertical normal effective stress); and δ is the friction angle between the
shaft and the soil.
For shafts in rock, the method presented in Section 8.3 can be used to calculate the
ultimate resistance pult.
(c) Validation of method

Zhang et al. (2000) verified the proposed method of analysis by comparing the results of
the proposed method with those obtained by other methods available in the literature and
with field test results including yielding.
Comparison with available elastic solution

The first verification concerns the elastic behavior of a shaft in a homogeneous half-space
with a constant modulus of elasticity E and Poisson’s ratio ν. In Figure 9.13, the results
obtained for a shaft having a length of 25 times its diameter (L/B=25) are shown as a
function of parameter Ep/G*, where Ep is the elastic modulus of the shaft and G* is the
modified shear modulus defined by
G*=G(1+3ν/4)
(9.61)
The displacement of the shaft head is expressed by the nondimensional parameter

uRG*/H, where R is the radius of the shaft. The general agreement between the results of
Verruijt and Kooijman (1989), those of Poulos (1971a) and those obtained by the
proposed
Fig. 9.13 Dimensionless displacement

of shaft head for soil or rock mass with
constant E (after Zhang et al., 2000).
method is good, although the displacements obtained by Verruijt and Kooijman (1989)
are greater than those predicted by other continuum methods at small Ep/G* values. The
dashed line in Figure 9.13 represents the results obtained by considering the soil or rock
mass as springs having a subgrade modulus equal to the modulus of elasticity in the
continuum model. The results for stiff shafts (large Ep/G* values) are remarkably good.
For flexible shafts (small Ep/G* values), however, this method results in an over-
estimation of the displacements.
As a second verification, the elastic behavior of a shaft in an elastic medium with a
linearly increasing modulus of elasticity is considered, assuming zero stiffness at the
ground surface, i.e., the shear modulus G of the soil or rock mass is expressed by
G=mz
(9.62)
The nondimensional displacement uR2m*/H of the shaft head is shown in Figure 9.14, as
a function of parameter Ep/m*R, where
m*=m(1+3ν/4)
(9.63)
Fig. 9.14 Dimensionless displacement

of shaft head for soil or rock mass with
G(E) increasing linearly with depth
As can be seen, the general agreement between the results of Verruijt and Kooijman
(1989), of Randolph (1981), of Banerjee and Davies (1978), and those obtained with the
proposed method is good, although the displacements of Randolph (1981) at small values
of Ep/m*R are slightly smaller than those predicted with the other continuum methods. As
in the case of a soil or rock mass with a constant modulus of elasticity, the spring
(subgrade) model overestimates the displacements for flexible shafts (small Ep/m*R
values), while the results for stiff shafts (large Ep/m*R values) correspond well to the
other predictions.
Comparison with field test results including yielding

The next two verifications compare the results obtained with the proposed method with
field test results including yielding at two sites by Frantzen and Stratten (1987). At each
site, two drilled shafts 0.22 m (8.6 in.) in diameter and 4.57 m (15 ft) long, were
constructed next to each other and were subjected to identical lateral loads. At the sandy
shale test site, the average unconfined compressive strength σc of the sandy shale is 3.26
MPa (34 TSF) and the average RQD is 55% (Frantzen & Stratten, 1987). To predict the
load-deflection response of the shaft, the deformation modulus Em and the ultimate
resistance pult of the rock mass have to be determined first. Since Em is not given in the
original report, the average deflection of the two shafts at the first recorded load 26.7 kN
[see Fig. 9.15(a)] is used to back-calculate the value of Em, assuming that the shafts
behave elastically at and below this load level. The back-calculated value of Em is 123
MPa. Next the ultimate resistance pult need be calculated. For sandy shale, the material
constant mi can be obtained from Table 4.5 as mi=12. Since GSI is not given in the
original report, it is approximately evaluated using the available information. Using
Bieniawski’s 1989 Rock Mass Rating (RMR) system, GSI can be evaluated by
GSI=RMR−5
(2.8)
To evaluate RMR, we have to know the unconfined compressive strength, RQD, the
spacing of discontinuities, the condition of discontinuities, the ground water conditions,
and the discontinuity orientations (see Table 9.1). When evaluating RMR in Equation
(2.8), a value of 15 is assigned to the groundwater rating and the adjustment for the
discontinuity orientation value is set to zero. Since we lack the information about the
spacing and condition of discontinuities for evaluating RMR, we assume the “average”
condition for both the spacing and condition of discontinuities. With this assumption,
RMR and thus GSI can be evaluated as shown in Table 9.1. Using the obtained GSI=
54>25, the material constants mb, s and a can be obtained from Equation (4.69),
respectively, as mb=2.321, s=0.00603 and a=0.5. Assuming that the effective unit weight
of the rock mass is 23 kN/m3, the ultimate resistance pult can be obtained, using the
method in Section 8.3.2, as shown in Figure 9.16(a). Using the deformation modulus and
ultimate resistance of the rock mass estimated above, the shaft head deflection at different
load levels can then be predicted. The comparison of the shaft head deflection obtained
from the field experiment and from the proposed method is shown in Figure 9.15(a). It
can be seen that the predicted deflections are in a reasonable agreement with those
measured. From the relative magnitudes of the predicted and measured deflections at
high load levels, we can clearly see that the shaft surface condition is between smooth
and rough.
Fig. 9.15 Comparison of test (Frantzen

& Stratten, 1987) and computed values
of shaft head deflection at sandy shale
and sandstone test sites (after Zhang et
al., 2000).
Fig. 9.16 Calculated ultimate lateral

resistance at sandy shale and sandstone
test sites (after Zhang et al., 2000).
Table 9.1 Values of estimated RMR89 and GSI

Rock Mass
Parameter Sandy Shale Sandstone
A1 Unconfined compressive strength 3.26 MPa 5.75 MPa
a)
Rating 1 2
A2 RQD 55% 45%
Ratinga) 13 8
A3 Spacing of discontinuities Assume: 200–600 mm
a)
Rating 10
A4 Condition of discontinuities Assume: Slightly rough surfaces, Separation < 1 mm,
Highly weathered walls
Ratinga) 20
A5 Ground water Hoek et al. (1995): Completely dry
a)
Rating 15
B Rating adjustment for discontinuity Hoek et al. (1995): very favorable
orientations
Ratinga) 0
RMR 59 55
GSI=RMR−5 54 50
a)
Rating value is assigned according to Table 2.12(a).
At the sandstone test site, the average unconfined compressive strength of the sandstone
is 5.75 MPa (60 TSF) and it has an average RQD of 45% (Frantzen & Stratten, 1987).
Using the average deflection of the two shafts at load 26.7 kN [see Fig. 9.15(b)], the
value of Em can be back-calculated as Em=170 MPa. GSI is evaluated as shown in Table
9.1. For sandstone, the material constant mi can be obtained from Table 4.5 as mi=19.
Using GSI=50>25, the material constants mb, s and a can then be obtained as mb=3.186,
s=0.00387 and a=0.5. Assuming again that the effective unit weight of the rock mass is
23 kN/m3, the ultimate resistance pult can be obtained as shown in Figure 9.16(b). The
shaft head deflection at different load levels is predicted as shown in Figure 9.15(b). It
can be seen that the predicted deflections are in a reasonable agreement with those
measured, the predicted results for rough socket conditions being closer to the measured
values than those for smooth socket conditions.
The predicted results (Fig. 9.15) show that the socket condition at the sandy shale test
site is somewhere in the middle between the smooth and rough conditions while the
socket condition at the sandstone test site is closer to the rough condition. This is as
expected because, for the same construction method, the socket in sandy shale should be
smoother than that in sandstone.
Example 9.2
In this example, the method of Zhang et al. (2000) is used to calculate the lateral
displacement of drilled shaft foundations of a planned cable stayed bridge. The geologic
profile is shown in Figure 9.17. The rock is a light brownish-gray to chocolate weathered
and unweathered, fine grained, plagioclase-quartz-biotite granofels and phyllite and it
includes thin beds of quartzite and fine grained schist. Since the geological conditions at
the east pier site are worse than those at the west pier site (see Fig. 9.17), only the east
shaft will be considered.
Due to the magnitude of the expected loads, the drilled shafts are proposed to be
socketed into the unweathered rock. Considering the influence of scour, the overburden
soil layer is ignored in the design and the shaft is assumed to be in a two-layer (weathered
and unweathered) rock mass system. The design parameters are summarized as follows:
1) diameter B and socket length L of the shaft: According to the construction methods
and the axial load design, B=3.0 m and L=4.0 m were selected
2) deformation properties of the shaft: Ep=30 GPa and νp=0.25
3) applied lateral load (i.e., the working load) H and M: The designers used many
different load combinations. For the most critical longitudinal loads acting in the East-
West direction, H=1.38 MN and M=51.1 MNm
4) Properties of the rock mass: RQD, σc, and Er of the weathered rock and the
unweathered rock are respectively as follows:
Weathered rock:
RQD=0 to 27 with an average=7, σc=6.9 MPa and Er=3.1 GPa.
Unweathered rock:
RQD=40 to 93 with an average=76, σc=67.6 MPa and Er=20.7
GPa.
The effective unit weight γ′ of both the weathered and unweathered rock masses
is assumed to be 13 kN/m3.
Solution:
Using the Em/Er—RQD relationship presented in Chapter 4, Em of the weathered and
unweathered rock masses can be obtained respectively as follows:
Weathered rock mass:
αE=0.0231(RQD)−1.32 (needs to be≥0.15)
=−1.158→0.15
Em=αEEr=0.15×3.1=0.46 GPa
Unweathered rock mass:

αE=0.0231(RQD)−1.32 (needs to be≥0.15)

=0.4356
Em=αEEr=0.4356×20.7=9.02 GPa
Fig. 9.17 Design example: Geological

profiles of East and West pier sites
For granofels and phyllite, mi=10. GSI is estimated to be 35 and 65 respectively for the
weathered rock mass and the unweathered rock mass. Using Equation (4.69), mb, s and a
can be obtained as follows:
Weathered rock mass (GSI=35>25): mb=0.9813, s=0.00073 and a=0.5
Unweathered rock mass (GSI=65>25): mb=2.865, s=0.0205 and a=0.5
The ultimate resistance pult is obtained, using the method in Section 8.3.2, as shown in
Figure 9.18.
The load-displacement and load-rotation relations of the shaft head, for both smooth
and rough shaft surface conditions, can be obtained as shown respectively in Figures 9.19
and 9.20. It is noted that at the working load H=1.38 MN and M=51.1 MNm, the shaft-
rock mass system yields slightly for the smooth shaft surface condition and acts
elastically for the rough shaft surface condition. The displacement and rotation of the
shaft head at the working load H=1.38 MN and M= 51.1 MNm are as follows:
For the smooth shaft surface condition:
u=3.703×10−3m, θ=1.713×10−3 rad
For the rough shaft surface condition:

u=3.568×10−3 m, θ=1.658×10−3 rad
The obtained displacement and rotation can then be checked against the allowable
design values.
Fig. 9.18 Calculated ultimate lateral

resistance of rock mass (after Zhang et
al., 2000).
Fig. 9.19 Predicted shaft head load-

displacement relations for the design
example with M/H= 37.0 m (after
Zhang et al., 2000).
Fig. 9.20 Predicted shaft head load-

rotation relations for the design
example with M/H=37.0 m (after
Zhang et al., 2000).
9.4 FINITE ELEMENT METHOD (FEM)
Randolph (1981) and Carter and Kulhawy (1992) used the finite element method (FEM)
to generate the parametric solutions for the load-displacement relations of laterally loaded
piles/shafts and, based on these solutions, they developed the closed-form expressions as
described in Section 9.3.1. The finite element method can also be used for analysis of
important structures and for study of the effect of important factors on the performance of
drilled shafts. Zhang (1999) used the finite element code ABAQUS (1998) to study the
effect of anisotropy of jointed rock mass on the deformation behavior of laterally loaded
drilled shafts in rock.
9.5 DISCRETE ELEMENT METHOD (DEM)
The discrete element method (DEM) is widely used in studying problems related to
fractured rock masses. Alfonsi et al. (1998) used the UDEC (Universal Distinct Element
Code) software to analyze drilled shafts in fractured rock masses. UDEC is a 2D discrete
element program specially designed to solve the discontinuous problem in which the
mechanical behavior of discontinuities can be directly simulated (Cundall, 1980).
Figure 9.21 shows the drilled shaft in a horizontal rock mass studied by Alfonsi et al.
(1998). There are two sets of discontinuities in the rock mass. The first set is vertical and
the second set has a dip angle of α (α=0° in Fig. 9.21). The intact rock elements are
assumed elastic with deformation modulus Er=10 GPa and Poisson’s ratio νr=0.25. The
discontinuities are assumed elasto-plastic with cj=0 kPa and . Assuming a
constant axial load of Q=2.5 MN, Alfonsi et al. (1998) obtained the lateral load-
displacement curves for three different values of α as shown in Figure 9.22. It can be seen
that α=0 (the second set of discontinuities are horizontal) provides much higher lateral
failure load than a =10° or 20° (the second set of discontinuities are inclined). This is
because the rock mass cannot fail by sliding along the discontinuities when α=0 (see Fig.
9.23).
Alfonsi et al. (1998) also analyzed drilled shafts in fractured rock slopes containing
two sets of discontinuities: the first set of discontinuities are persistent and have a dip
angle a, and the second set of discontinuities are non-persistent and perpendicular to the
first set. Keeping a constant axial load of Q=55 MN and increasing the lateral load H,
Alfonsi et al. (1998) obtained the failure modes for three different discontinuity
orientations (expressed by α) as shown in Figure 9.24. When the persistent discontinuity
set dips down with α= 60° [see Fig. 9.24(a)], no clear rupture line is formed and the
maximum lateral load obtained is 4.5 MN. When the persistent discontinuity set is
horizontal (α=0°) [see Fig. 9.24(b)], sliding occurs along the persistent discontinuities
and the maximum lateral load obtained is only 3.5 MN. When the persistent discontinuity
set dips up with α=30° [see Fig. 9.24(c)], a stair-shape rupture line is formed and the
maximum lateral load obtained is very high, i.e., 18.5 MN.
Fig. 9.21 Drilled shaft in rock analyzed

using UDEC (after Alfonsi et al.,
1998).
Fig. 9.22 Lateral load-displacement

curves obtained using UDEC (after
Alfonsi et al., 1998).
Fig. 9.23 Influence of discontinuity

orientation on the failure pattern of
rock mass (after Alfonsi et al., 1998).
Fig. 9.24 Failure modes for three

different discontinuity orientations
(after Alfonsi et al., 1998).
It should be noted that the results shown in Figures 9.22 to 9.24 are from the 2D analyses.
The actual 3D performance of drilled shafts in fractured rock masses will be affected by
the third discontinuity set (Alfonsi et al., 1998, 1999). As shown in Figure 9.25, the rock
mass volume moved depends on the pattern of the third discontinuity set.
9.6 DRILLED SHAFT GROUPS
Numerous methods exist for analyzing laterally loaded pile groups in soil, some of which
can be applied to drilled shaft groups in rock and are briefly described in the following.
9.6.1 Deflection ratio approach

The deflection ratio approach for calculating the deflection and rotation of a pile group in
soil has been described by Poulos and Davis (1980). The approach involves superposition
of lateral interaction factors, and is similar in principle to the analysis described in
Chapter 7 for axially loaded pile groups.
For pinned-head piles, the group deflection can be expressed as follows

uG=uavRu
(9.64)
where uG is the horizontal deflection of the pile group; uav is the horizontal deflection of a
single pile at the average load level of a pile in the pile group; and Ru is the group
deflection ratio for a pinned-head pile group.
For fixed-head piles, the group deflection is
uG=uavRF
(9.65)
where uav is as above; and RF is the group deflection ratio for a fixed-head pile group.
For piles which are rigidly attached to the pile cap, but the pile cap can rotate, the
response of the pile group is dependent on both the lateral and axial characteristics of the
piles. However, for such groups, the lateral group deflection is found to be only slightly
greater than that for a fixed-head group, so that, for practical purposes, Equation (9.65)
may be used.
The values of Ru and RF for different soil profiles can be found in Poulos (1979) for a
variety of group configurations, pile spacings and relative stiffnesses. These values may
be used with the single shaft deflection computed in Section 9.3 to estimate the deflection
of a drilled shaft group in rock. However, since the values of Ru and RF are obtained
specifically for piles in soil, they can only be used in the very preliminary design. For the
final design of major projects, it is desirable, when feasible, to conduct lateral load tests
on groups of two or more drilled shafts in rock in order to confirm the Ru and RF values of
Poulos (1979) or to derive new, site-specific values.
Fig. 9.25 Influence of discontinuity

pattern on the rock mass volume
moved (after Alfonsi et al., 1998).
9.6.2 p-y curve approach

The p-y curve approach can be used to estimate group action by introducing a “p-
multiplier” suggested by Brown et al. (1988) to modify the p-y curve for a single drilled
shaft. That is, all of the values of rock resistance p are multiplied by a factor that is less
than 1, the p-multiplier, depending upon the location of the shaft within the group and the
spacing of the shafts within the group. That is, all along the p-y curve (Fig. 9.26):
pgroup=ρPsingle
(9.64)
where ρ is the p-multiplier. This factor reflects a dominant physical situation that
develops within a laterally loaded group of drilled shafts: The shafts in the leading row
push into the rock in front of the group. The rock reacting against any drilled shaft in this
“front row” is relatively unaffected by the presence of other drilled shafts in the group
and only a minor adjustment needs to be made to the p-y curves. However, the shafts in
the rows that “trail” the front row obtain resistance from rock that is pushed by the shafts
into the voids left by the forward movement of the shafts in front of them. This
phenomenon causes the value of rock resistance p on a p-y curve to be reduced at any
given value of lateral deflection y
Fig. 9.26 Modification of p-y curves

using the p-multiplier ρ (after Brown et
al., 1988).
relative to the values that would exist if the drilled shafts in the forward row were not
there. In addition, the presence of all the shafts in the group produces a mass movement
of the rock surrounding the shafts in the group, which reduces the p-value for a given
displacement y to varying degrees for all drilled shafts in the group (O’Neill & Reese,
1999).
Table 9.2 lists the most commonly used p-multiplier recommendations at 3 diameter
center-to-center (3B) spacing. Rollins et al. (1998) recommend that the p-multipliers
increase to unity at a center-to-center spacing of 6B. FHWA (1996b) recommend using
the p-multipliers presented in Table 9.2 for all center-to-center spacings.
It should be noted that the p-multipliers that have been developed are primarily for
driven piles in soil. To apply the p-multipliers to drilled shafts in rock, the following two
issues need be considered:
1. The difference between soil and rock in their performance
2. The stress relief around existing drilled shafts when new drilled shafts are installed
adjacent to the existing drilled shafts.
Therefore, it is desirable, when feasible, to conduct lateral load tests on groups of two or
more drilled shafts in rock for major projects in order to confirm the p-multipliers in
Table 9.2 or to derive new, site-specific values.
Table 9.2 Common p-multiplier recommendations
at 3 diameter center-to-center spacing.
Recommendation FHWA (1996b) Rollins et al. (1998)
Lead row 0.8 0.6
2nd row 0.4 0.4
3rd row 0.3 0.4
Software that uses p-y curves to analyze laterally loaded drilled shafts allow the user to
input values of the p-multiplier, based on the recommendations in Table 9.2 or based on
other information, such as site-specific load tests. When using a single-shaft computer
code to analyze a group of identical, vertical, laterally loaded drilled shafts subjected to a
shear load at the elevation of the shaft heads and a concentric axial load, it is advisable to
use the lateral displacement rather than the lateral load (applied shear at the shaft head) as
a head boundary condition (O’Neill & Reese, 1999). The head restraint condition (free,
fixed or intermediate restraint) is used as the other head boundary condition, depending
upon how the shaft is connected to the cap. A typical front row shaft is analyzed using the
p-multiplier for the front row. This analysis gives the head shear, moment and rotation, as
well as the deflected shape of the shaft and the shear and moment distribution along the
shaft for the front-row shafts. This analysis is repeated for a typical drilled shaft on a
trailing (back) row applying the same value of head deflection and using the value of p-
multiplier for shafts on back rows to modify the p-y curves. Similar output is obtained.
The shear load that must have been applied to the group to produce the assumed lateral
head deflection is then equal to the head shear on a front row shaft times the number of
shafts on the front row plus the head shear on a back row shaft times the number of shafts
in the group that are not on the front row. If this shear is not equal to the applied shear, a
different head displacement is selected and the process is repeated until the computed
head shears of all of the shafts in the group sum to the applied group shear. The moment
and shear distributions for the shafts in the front row will be different from those for the
shafts not on the front row; therefore, different steel schedules will often be appropriate
among the shafts in the various rows within the group (O’Neill & Reese, 1999).
9.6.3 Finite element method (FEM)

Advances in computer technology have made it possible to analyze laterally loaded
drilled shaft groups using 3D finite element (FE) models. p-y curves or sophisticated
constitutive relations can be used to represent the rock behavior in the 3D FE analyses. It
should be noted that, however, it is often difficult if not impossible in design practice to
obtain the parameters in the sophisticated constitutive relations.
10
Stability of drilled shaft foundations in rock
10.1 INTRODUCTION
Drilled shaft are frequently installed in rock slopes, for example, the foundations for
power poles and bridges. In many cases, drilled shafts are also used to stabilize rock
slopes. Because rock masses often contain discontinuities, it is important to check the
stability of rock blocks or wedges formed by the discontinuities.
Blocks formed by discontinuities may fail in different modes (Hoek & Bray, 1981).
Figures 10.1(a) to (c) show respectively the planar sliding failure on a single
discontinuity, the wedge sliding failure on two intersecting discontinuities, and the
toppling failure of toppling blocks. In weathered or highly fractured rock masses, the
failure surface is less controlled by single through-going discontinuities. In this case, an
approximately circular failure surface may develop in a similar manner to failures in soil
[Fig. 10.1(d)]. To be general, the drilled shafts in Figure 10.1 are subject to not only axial
load but also lateral load. The failure surface may cut through the shaft or pass through
below the shaft base.
10.2 PLANAR SLIDING FAILURE
When discontinuities are approximately parallel to and dip out of the slope face, a planar
sliding failure may be formed along the discontinuities as shown in Figure 10.2. The
stability of the block is defined by the relative magnitude of two forces acting parallel to
the potential sliding surface: the driving force F acting down the surface, and the resisting
force R acting up the surface. The factor of safety (FS) of the block is simply defined by
(10.1)
For potential sliding along a persistent discontinuity as shown in Figure 10.2(a), the total
driving force F acting down the sliding surface can be calculated by:
F=Wsinα+Qsinα+Hcosα
(10.2)
Fig. 10.1 Failure modes of rock slopes:

(a) Planar sliding failure on a single
discontinuity; (b) Wedge sliding
failure on two intersecting
discontinuities; (c) Toppling failure of
steeply dipping slabs; and (d)
Approximately circular failure in
weathered or highly fractured rock
masses.
where α is the angle between the discontinuity and the horizontal plane; Q and H are
respectively the axial and lateral loads acting on the shaft head; and W is the total weight
of the block including the drilled shaft. The resisting force R acting up the potential
sliding surface is
Stability of drilled shaft foundations in rock 351
(10.3)
where cj and are respectively the cohesion and internal friction angle of the
discontinuity; and l and b are respectively the length and width of the sliding surface. The
Fig. 10.2 Potential sliding along: (a) A

persistent discontinuity; and (b) Non-
persistent discontinuities.
selection of b is critically important in analyzing the stability of the block and will
significantly affect the FS calculated. If there is a row of drilled shafts in the longitudinal
direction of the slope at spacing s, b can be simply selected equal to s.
If there is no lateral load acting on the drilled shaft head and the cohesion of the
discontinuity is zero, the FS is simply given by
(10.4)
that is, the limiting condition occurs when the dip of the sliding surface equals the friction
angle of the discontinuity.
For non-persistent discontinuities as shown in Figure 10.2(b), the failure surface will
also pass through rock bridges. For this case, the method of Einstein et al. (1983) can be
extended to calculate the stability of the block. The principle of this method is illustrated
in a simplified form in Figure 10.3: the slope overlying the failure path is partitioned into
a series of vertical slices, bounded at their bottom end by discontinuities or intact rock.
The total driving force F can be calculated by summing slice contributions and the forces
acting on the shaft head:
(10.5)
where α is the angle between the discontinuity and the horizontal plane; Q and H are
respectively the axial and lateral loads acting on the shaft head; and Wi is the weight of
the ith slice. For the slices containing the drilled shaft, it is important to include the
weight of the shaft.
Fig. 10.3 Dividing the potential sliding

wedge into slices to calculate total
driving force F and total resisting force
R (Modified from Einstein et al.,
1983).
The total resisting force R is calculated by summing slice contributions:
(10.6)
where Ri is the shear resistance mobilized by the portion of path underlying that slice.
The ith portion of the path may be jointed or consist of intact rock. The calculation of Ri
for those two cases is described in the following.
For the ith portion of the path along a discontinuity, Ri can be simply calculated by
(10.7)
where cj and are respectively the cohesion and internal friction angle of the
discontinuity; li and b are respectively the length and width of the ith portion of the path
along the discontinuity; and σai is the effective normal stress on the discontinuity plane,
which can be simply calculated by
(10.8a)
(10.8b)
For the ith portion of the path through the intact rock, Ri need be calculated in two ways
based on the failure modes of the rock bridges. For in-plane or low-angle out-of-
plane transitions (βi<θt, see Fig. 10.4 for βi and Fig. 3.7 for θt), the intact-rock
resistance Ri can be calculated by
Ri=τaidib
(10.9)
where di is the “in-plane length” of the rock bridge (Fig. 10.4); and τai is the peak shear
stress mobilized in the direction of discontinuities which can be obtained by
(10.10)
where σt is the tensile strength of the intact rock; and σai is the effective normal stress on
the discontinuity plane, which can be simply calculated using Equation (10.8).
For high angle transitions (βi>θt, see Fig. 10.4 for βi and Fig. 3.7 for θt), the intactrock
resistance Rican be calculated by
Ri=σtXib
(10.11)
where Xi is the distance between discontinuity planes that define the bridge (Fig. 10.4).
In Figure 10.2, the drilled shaft is above the potential sliding surface. If the drilled
shaft penetrates through the sliding surface as shown in Figure 10.5, the contribution of
the drilled shaft to the stability of the block need be considered in the stability analysis.
One simple way is to include the shear resistance of the drilled shaft in the resisting force
R. It is important to note that the bending and shear capacity of the drilled shaft should be
checked so that no structural failure will occur for the drilled shaft itself. The lateral load
capacity provided by the portion of the shaft below the sliding surface should also be
checked.
Fig. 10.4 Definition of βi, Xi and di

(Modified from Einstein et al., 1983).
Fig. 10.5 Drilled shaft penetrating

through the sliding surface.
Example 10.1
A row of drilled shafts of diameter 0.61 m is to be installed in a rock slope (Fig. 10.6).
The properties of the drilled shafts and the rock mass are as follows:
Spacing of the drilled shafts in the longitudinal direction, b=3.0 m
Shear strength properties of the discontinuity, cj=0 and

Unit weight of the rock mass, γ=23 kN/m3
Loads at the shaft head, Q=2,500 kN, H=100 kN
Length of the shaft above the rock surface, e=4.0 m
Length of the sliding surface along the discontinuity, l=14.4 m
Other properties are shown in the figure
Evaluate the stability of the rock slope along the discontinuity.
Solution:
The gross cross-sectional area of the drilled shaft is Ag=0.257π×0.612=0.292 m2.
Assuming that the drilled shaft has the same unit weight as the rock mass the weight
of the rock block including the drilled shaft can be estimated as follows:
W=[0.5×14.4×14.4×0.5×tan(30°)×3.0+0.292×4.0]×23=2,092 kN
Using equation (10.2), the total driving force acting down the discontinuity surface is
F=2,092×sin(30°)+2,500×sin(30°)+100×cos(30°)=2,383 kN
Using equation (10.3), the total resisting force acting up the discontinuity surface is
R=[2,092×cos(30°)+2,500×cos(30°)−100×sin(30°)]×tan(31°)=2,359 kN
Fig. 10.6 Stability analysis of a rock

slope containing drilled shafts.
Using Equation (10.1), the factor of safety of the rock slope along the discontinuity is
FS=2,359/2,383=0.99
which is below 1, meaning the rock slope will slide along the discontinuity. It is noted
that the factor of safety of the rock slope without the drilled shafts is
which is above 1, meaning the rock slope is stable. So the installation of the drilled shafts
without passing through the discontinuity plane decreases the factor of safety of the rock
slope because of the lateral load at the shaft head.
If the drilled shafts pass through the discontinuity plane, the shear resistance of the
drilled shafts will increase the stability of the rock slope. Assuming a concrete strength of
f′c=28,000 kPa, the limiting concrete shear stress can be estimated from Equation (8.2) as
νc=2.63×(1+25,000/13,780/0.292)×(28,000)0.5=714 kPa
The area of the drilled shaft resisting shear along the discontinuity plane is estimated as
Av=0.95Ag/cos(α)=0.95×0.292/cos(30°)−0.32 m2
So the shear resistance of the drilled shaft along the discontinuity plane is approximately
Vu=νcAv=714×0.32=229 kN
The factor of safety of the rock slope with drilled shafts passing through the discontinuity
is
FS=(2,359+229)/2,383=1.09
which is above 1.
It is important that the drilled shafts be extended long enough below the sliding plane
to provide sufficient lateral load capacity. The structural resistance of the drilled shafts
should also be checked.
10.3 WEDGE SLIDING FAILURE
A wedge failure is formed by two intersecting discontinuities dipping out of but aligned
at an oblique angle to the slope face, the slope face and the upper slope surface [Fig.
10.1(b)]. The general failure mode is by sliding parallel to the line of intersection of the
two discontinuities. The method of stability analysis of wedge blocks follows the same
principles as that of the planar blocks, except that it is necessary to resolve forces on both
of the sliding planes. For the detailed procedure for calculating the factor of safety of
three-dimensional wedge blocks, the reader can refer to Hoek and Bray (1981) and
Wyllie (1999). In the analysis, it is important to include the influence of the drilled shaft,
including the size and location of the shaft and the applied axial and lateral loads at the
shaft head. The drilled shaft may cause a stable wedge block to slide due to the loads at
the shaft head or stabilize an unstable wedge block by extending beyond the potential
sliding planes.
10.4 TOPPLING FAILURE
Toppling failure may occur where discontinuities dip into the face and form either a
single block, or series of slabs, such that the center of gravity of the block falls outside
the base (Wyllie, 1999). The analysis of toppling failure of drilled shaft foundations can
be conducted by examining the stability conditions of each block in turn starting at the
top of the slope, following the procedure of Goodman and Bray (1976) and Wyllie
(1999). In the analysis, it is important to include the influence of the drilled shaft,
including the size and location of the shaft and the applied axial and lateral loads at the
shaft head. The drilled shaft may cause a stable block to topple or stabilize an unstable
block, depending on the size and location of the shaft and the applied axial and lateral
loads at the shaft head.
10.5 CIRCULAR FAILURE
An approximately circular failure may occur in weathered or highly fractured rock

masses, where the failure surface is less controlled by single through-going
discontinuities but passes partially through intact rock and partially along existing
discontinuities. The analysis of circular failure surfaces in rock may follow the same
methodology for analyzing circular failure surfaces in soil. Slope stability analysis
programs such as UTEXAS3 (Wright, 1991) and XSTABL (Sharma, 1991) can be
readily used in the analysis of circular failure surfaces in rock. Again, the influence of the
drilled shaft, including the size and location of the shaft and the applied axial and lateral
loads at the shaft head, need be considered in the stability analysis.
11
Drilled shafts in karstic formations
11.1 INTRODUCTION
Drilled shafts are frequently used in karstic formations. The challenges to using drilled
shafts in karstic formations involve the highly irregular nature of the rock-overburden
interface and the cavities in the bearing rock. The erratic nature of the bearing rock
surface may require drilled shafts of different length be used. Two “depth of bedrock”
borings cannot be simply connected by a straight line when inferring the rock surface
from borings. An existing cavity underlying a drilled shaft may collapse after the building
is in service. In blanketed active karst, new sinkholes may form and lead to collapse of
drilled shafts.
Drilled shafts in karstic formations may fail in different modes as shown in Figure 1.3.
Since large structural load is supported by each drilled shaflt the failure of any one shaft
may cause critical damage to the entire structure. Therefore, special care must be taken
for drilled shafts in karstic formations.
11.2 CHARACTERISTICS OF KARSTIC FORMATIONS
Approximately one-fourth of the earth’s land surface is underlain by rocks which are
susceptible to solutioning activity (Cooper & Ballard, 1988). These rocks include
limestone, dolomite, gypsum, anhydrite, and salt (halite) formations. Karst terrain
develops through continuous erosion of soluble rock minerals over a long period of time.
When rainwater falls onto the ground surface and percolates downward into cracks
and fissures, it gradually dissolves the rock and leaves insoluble materials such as chert
and clay behind. Since the weathering resistance of rocks is variable, areas of least
weathering develop high rock pinnacles and areas of severe weathering develop deep
slots and cavities. This results in an extremely irregular rock and overburden interface
such as that shown in Figure 11.1.
Rock solution can also result in enlargement of interparticle porosity, decreasing the
rock strength and increasing the compressibility. Continuing solution enlargement of the
interparticle porosity can result in coalescence of voids to form cavities.
Fig. 11.1 Bedrock surface in a thinly

mantled karst terrane, West Central
Florida (after Wilson & Beck, 1988).
When a rock cavity enlarges due to rock solution, the shear and tensile stresses in the
cavity roof and the compressive stresses in the cavity walls increase, with the maximum
shear stresses between. Continuing enlargement of a rock cavity may result in the
collapse of the cavity roof.
11.3 INVESTIGATION OF KARSTIC FORMATIONS
Because of the peculiar nature of karstic formations, it is extremely important to conduct

site investigations to identity the degree of dissolution and the pattern and extent of
specific hazards, such as cavities and sinkholes, erosion domes and the potential for their
further development.
Site investigations in karstic formations also include the three main stages as described
in Chapter 5: (1) preliminary investigations, (2) detailed investigations, and (3) review
during construction and monitoring. The following briefly describes the main points
related to site investigations in karstic formations.
Preliminary investigations begin with an intensive review of existing information. The
fundamental data include geological maps and reports of the area to identify the
underlying rock and, particularly, any carbonate rocks and their approximate geographic
boundaries. In areas of significant cave development, local, regional and national cave
exploration groups have prepared and compiled descriptions, detailed maps, and reports
on specific caves. Table 11.1 is a summary of major karst areas in the United States.
The second tool of preliminary investigations is remote sensing using air photographs,
infrared imagery, and side-looking radar. For most of the United Sates, air photographs
are available from the U.S. Geological Survey, the U.S. Forest Service, and the U.S.
Department of Agriculture. Air photographs can be used to detect karstic terrain which is
shown by such topographic features as basin-studded plains, narrow U-shaped valleys
with vertical sides, rolling topography, and scalloped effect around river
Drilled shafts in karstic formations 361
Table 11.1 Summary of major karst areas in the

United States (after ASCE, 1996).
Karst Area Location Characteristics
Southeastern South Carolina, Rolling, dissected plain, shallow dolines, few caves;
coastal plain Georgia tertiary limestone generally covered by thin deposits of
sand and silt.
Florida Florida, southern Level to rolling plain; tertiary, flat-lying limestone;
Georgia numerous dolines, commonly with ponds; large springs;
moderate sized caves, many water filled.
Appalachian New York, Vermont, Valleys, ridges, and plateau fronts formed south of
south to northern Palaeozoic limestones, strongly folded in eastern part;
Alabama numerous large caves, dolines, karst valleys, and deep
shafts; extensive areas of karren.
Highland Rim Central Kentucky, Highly dissected plateau with Carboniferous, flat-lying
Tennessee, northern limestone; numerous large caves, karren, large dolines
Georgia and uvala.
Lexington North-central Rolling plain, gently arched; lower Palaeozoic
Nashville Kentucky, central limestone; a few caves, numerous rounded shallow
Tennessee, south dolines.
eastern Indiana
Mammoth West-central, Rolling plain and low plateau; flat-lying Carboniferous
Cave southwestern Kentucky, rocks; numerous dolines, uvala and collapse sinks; very
Pennyroyal southern Indiana large caves, karren developed locally, complex
Plain subterranean drainage, numerous large “disappearing”
streams.
Ozarks Southern Missouri, Dissected low plateau and plain; broadly arched Lower
northern Arkansas Palaeozoic limestones and dolomites; numerous
moderatesized caves, dolines, very large springs; similar
but less extensive karst in Wisconsin, lowa, and
northern Illinois.
Canadian Western Oklahoma, Dissected plain, small caves and dolines in
River northern Texas Carboniferous gypsum.
Pecos Valley Western Texas, Moderately dissected low plateau and plains; flat-lying
southeastern New to tilted Upper Palaeozoic limestones with large caves,
Mexico dolines, and fissures; sparse vegetation; some gypsum
karst with dolines.
Edwards Southwestern Texas High plateau, flat-lying Cretaceous limestone; deep
Plateau shafts, moderate-sized caves, dolines; sparse vegetation.
Black Hills Western South Dakota Highly dissected ridges; folded (domed) Palaeozoic
limestone; moderate-sized caves, some karren and
dolines.
Kaibab Northern Arizona Partially dissected plateau, flat-lying Carboniferous
limestones; shallow dolines, some with ponds; few
moderate-sized caves.
Western Wyoming, north Isolated small areas, primarily on tops and flanks of
mountains western Utah, Nevada, ridges, and some area in valleys; primarily in folded and
western Montana, tilted Palaeozoic and Mesozoic limestone; large caves,
Idaho, Washington, some with great vertical extent, in Wyoming, Utah,
Oregon, California Montana, and Nevada; small to moderate-sized caves
elsewhere; dolines and shafts present; karren developed
locally.
systems, with streams entrenched in bedrock on rectangular patterns. It is often useful

to examine photographs over an extended time period which may show, for example, the
progressive development of solutioning, or that sinkholes may have been obscured by
human activities.
Site reconnaissance is another important part of preliminary investigations and
includes examining the area for verification of changes in previously observed or
photographed features. Site reconnaissance should also be made to examine suspicious
terrain details that are difficult to see from the air photographs because of tree and
vegetation cover as well as overhangs and other obstructions.
Geophysical methods, such as ground penetration radar (GPR), seismic survey and
electrical resistivity, are widely used in site investigations in karstic terrains. As
geophysical survey becomes quicker and cheaper, geophysical work is conducted in the
stage of preliminary investigation as well as in the stage of detailed investigation. The
selection of the most appropriate technique(s) for a site will depend on the particular site
conditions. For example, GPR has been successfully used in Florida where the cavities
are overlain by sand (Benson, 1984), but it is less successful in Pennsylvania where
overlain is clayey soil with a high moisture content (Wyllie, 1999). The reliability of
geophysics to detect cavities and predict their shape and size is limited because cavities
may have irregular shapes and be filled with different materials such as air, water, clay
and boulders. For this reason drilling is often carried out at the stage of detailed
investigation.
Because of the irregular rock and overburden interface and the cavities in rock, the
number of borings required per unit of site area in karstic formations is usually much
larger than for site investigations in other formations. In cases such as that the rock
masses contain steep or vertical discontinuities, a few inclined borings may be required.
Table 11.2 lists the boring and sampling techniques that may be used for foundation
exploration in karstic formations.
11.4 CONSIDERATIONS IN THE DESIGN AND CONSTRUCTION

OF DRILLED SHAFTS IN KARSTIC FORMATIONS
Design of drilled shafts in karstic formations is generally based on end bearing resistance
in the hard rock (Brown, 1990). Because it is difficult to form a clean socket in massive
rock, the side friction of the rock socket is usually ignored or assigned an extremely
conservative value. In areas where fault zones or other geological features have produced
deep slots that have virtually no sound rock, drilled shafts may be designed for side
friction only (Brown, 1990).
In karstic formations, it is essential that the foundation rock below the bottom of each
drilled shaft be explored for defects. The defects include cavities that could allow the
rock below the drilled shaft to crush or break under the future foundation load or clay-
filled seams that would allow the foundation to subside as the clay consolidates or
extrudes outward under the concentrated foundation load. Inspection of the bearing rock
below the bottom of a drilled shaft is performed both by inspecting the bearing surface
and by drilling one or more probe holes to a depth of at least two shaft diameters below
the bearing surface (Fig. 11.2). The walls of the probe holes can be probed to find any
small open seams or cavities that would compress or allow the rock to fracture under load
transferred by the shaft. This is done with a steel tube or rod fitted with a small,
horizontal wedge-shaped tip that is pressed against the hole wall as it is lowered into, or
pulled out of, the probe hole. The tip can find open seams less than 1 mm thick and clay
seams as thin as 1 to 2 mm.
Table 11.2 Boring and sampling techniques for
foundation exploration in karstic formations (after
Sowers, 1996).
• Percussion drilled holes to identify soil-rock interface. The observed drilling rate is an indicator
of rock hardness and rock discontinuities such as fissures and voids. This requires recording the
rate of penetration for short intervals of drilling: minutes per foot or per meter, as well as visual
examination of the drill cuttings. However, it is difficult to differentiate between a large boulder,
a pinnacle, and the upper surface of continuous rock.
• Test borings with intact split-spoon samples and Standard Penetration Tests (SPT) in soillike
materials, particularly in the soft zone immediately above rock and in the soil in cavities within
the rock after drilling into the rock. The boreholes are made by augers or rotary cutters using air
or drilling fluid to remove the cuttings. Any loss of drilling fluid is measured as an indication of
the size and continuity of rock fissures and cavities. Laboratory tests of the samples provide data
for accurate classification of the soils and for estimating some of the engineering characteristics
such as hydraulic conductivity and response to loading.
• Undisturbed sampling of the stiff overburden at representative intervals and, if possible, of the
soft soil overlying the rock and filling slots and cavities. The size of the sample tube is
determined by the boring diameter. Laboratory tests of the samples provide quantitative data for
engineering analyses of the soil hydraulic conductivity and response to loading.
• Cone penetration tests in soil, particularly in the soft soil zone and in soil in the rock (after core
drilling or percussion drilling exposes the soil seams). The usual cone point and sleeve resistance
can be supplemented by pore water pressure sensors using a piezocone.
• Core boring preferably with triple tube diamond bits in rock. When the rock is so weak or
closely fractured that the core recovery is less than approximately 90%, larger diameter cores, 4
to 6 in (100 to 150 mm), are preferred.
• Oriented core drilling to determine the dip and dip azimuth of the strata and of fissures.
• Large diameter drill holes that permit human access to examine the exposed materials directly,
particularly at the soil-rock interface. The minimum diameter is approximately 30 in (760 mm);
holes of 36 in (900 mm) or larger are preferred. Direct access requires casing in the hole for
safety, which means alternating drilling, setting casing, and making the observations. This is
often impractical below the groundwater level.
• Test pits or test trenches to check the rock surface, exposing both slots and pinnacles. This
requires either flat slopes or bracing to prevent cave-ins and to provide safety. It is often
impractical below the groundwater level despite heroic pumping to dewater the bottom.
Moreover, pumping could trigger sinkhole activity. Pits and trenches make it possible to view
the stratification, the orientation of fissures, as well as the geometry of the soil-rock interface in
three dimensions.
• Borehole photography or video imaging of the borehole walls.
Fig. 11.2 Finding defects in the rock

below the bottom of a drilled shaft
with the aid of small diameter probe
holes and a hand probe: (a) Probe
holes drilled in the rock below the
bottom of a drilled shaft; and (b) Probe
rod for finding rock defects on the
walls of probe holes.
Fig. 11.3 Drilled shaft bearing over

vertical soil filled seam (after Brown,
1990).
If a vertical discontinuity or seam exists as shown in Figure 11.3, the rock that is present
over the base of the shaft need be probed. If sound rock exists except for a small area
(rock coverage over 75% of the shaft base), the hole may be accepted as adequate if the
resulting bearing pressures are not excessive (Brown, 1990). If possible, the soil in the
seam can be excavated and the seam backfilled with concrete so that higher end bearing
capacity can be achieved. For deep shafts drilled below the groundwater table, this
practice is not recommended because of the possibility of large and uncontrolled seepage
into the shaft through such an excavated seam.
If the soil filled seam is located at one side of the drilled shaft and the rock below the
bottom of the shaft contains non-vertical discontinuities as shown in Figure 11.4, rock
anchors may be used to provide continuity and load transfer across the discontinuities
(Brown, 1990; Goodman, 1993; Sowers, 1996).
If the rock is of insufficient quality to provide the required end bearing capacity, the
shaft can be extended deeper so that the foundation load is partially transferred to the
rock by side shear. Figure 11.5 shows that extension of the shaft reduces the vertical
stress in the underlying rock by transferring much of the load from the shaft to the rock
by side shear (Sowers, 1996). If a deep, near-vertical soil filled seam is directly under the
bottom of the shaft, it is also possible to increase the load capacity of the shaft by
extending the shaft deeper, which is true especially when the rock on both sides of the
seam is intact and sound (Fig. 11.6).
If there exist caves below the shaft, different remedial measures can be taken based on
the size and location of the caves and the condition of the rock. If the roof of the cave is
thin, the drilled shaft can be extended through the roof and cave and into the sound rock
in the cave floor (Fig. 11.7). Casing is required for the shaft through the cave. If the roof
of the cave is thick and the rock is sound, the drilled shaft can be used as a side shear
only shaft [Fig. 11.8(a)]. In some cases, drilling may break through into the cave at the
depth where drilling would normally terminate [Fig. 11.8(b)]. To use the hole for a shaft
without extending it through the cave and into the rock in the cave floor, a wood plug can
be inserted into the bottom of the hole before casting concrete. Obviously, the shaft with
the wood plug will only provide side shear resistance. If the cave below the shaft is small,
drilling can be extended through into the cave so that the cave can be filled with concrete.
This will minimize subsidence and prevent catastrophic collapse.
Fig. 11.4 Rock anchors for drilled shaft

bearing over rock containing
discontinuities (after Brown, 1990).
Fig. 11.5 Extending drilled shaft to

reduce the vertical stress in the
underlying rock by transferring load
from shaft to rock by side shear: (a)
Underlying rock cannot provide
required end bearing capacity; and (b)
Shaft extension and assumed average
vertical rock stress distribution
(modified from Sowers, 1996).
Fig. 11.6 Remedial measures for

drilled shaft with a deep, near-vertical
soil filled seam directly under its
bottom: (a) Drilled shaft with a deep,
near-vertical soil filled seam directly
under its bottom; and (b) Shaft

extension and added, deeper probe
holes (modified from Sowers, 1996).
Fig. 11.7 Extending drilled shaft

through the roof and cave to the sound
rock in the cave floor (modified from
Sowers, 1996).
Fig. 11.8 (a) Side shear only shaft with

a cave below it; and (b) Using wood
plug at the bottom of the hole in the
case that drilling breaks through into
the cave.
11.5 AN EXAMPLE OF DRILLED SHAFT FOUNDATIONS IN

KARSTIC FORMATIONS
This example is from Erwin and Brown (1988). It shows the problems and the
corresponding solutions to them for drilled shaft foundations in karstic formations.
Since 1962, active limestone sinkholes had been causing problems and affecting the
operation of the access railroad to the Military Ocean Terminals, Sunny Point, North
Carolina (MOTSU). Many springs and sinkholes developed in areas close to the MOTSU
access railroad (Fig. 11.9). Two sinkholes that developed in October 1984 following
extensive rainfall caused all traffic to be stopped. To reactivate railroad traffic, a 450 ft
(137.2 m) bypass alignment was selected based on a ground penetration radar (GPR)
survey. An 8,500 lb (3,855 kg) concrete weight was dropped from a 30 ft (9.14 m) height
every 15 ft (4.57 m) along the alignment. No sinkholes were activated by dropping the
weight. Two potential sinkholes identified by the GPR survey were grouted. The two
sinkholes were encircled with 6,500 ft3 (184.0 m3) of sanded grout, excavated 8 to 10 ft
(2.44 to 3.05 m) deep, compacted by dropping concrete weight, and backfilled with
compacted sand.
The stratigraphy along the MOTSU access railroad consists of the following materials
in descending order:
1. Sand and silt of Pleistocene age that are loose and noncemented: 10 to 40 ft (3.05 to
12.2 m) thick.
2. Silt, clayey sand, shell hash, and shell layers of the early Pleistocene/Pliocene age
Waccamaw formation up to 30 ft (9.14 m) thick.
3. Limestone of the Castle Hayne formation of upper to middle Eocene age: 5 to 33 ft
(1.52 to 10.1 m) thick.
4. Limestone, sandstone, sand, and silt of the Cretaceous age Pee Dee formation: 5 to 10
ft (1.52 to 3.05 m) thick.
Extensive subsurface investigations indicated that there is a set of southeast-northwest
trending joints that appear to connect to a lake upstream of the railroad.
In January 1985, a 4,055 ft (1.24 km) land bridge was selected as a permanent solution
to the sinkhole problems for the MOTSU access railroad. Drilled shafts socketed into
rock were selected to support the bridge. For areas of known sinkhole activity, it was
assumed that 20 ft (6.10 m) of overburden could be lost in two consecutive bents from
sinkhole formation (see Fig. 11.10). For areas of no known foundation problems, it was
assumed that no more than 15 ft (4.57 m) of overburden could be lost. In the Allen Creek
area, it was assumed that there was no overburden available for lateral support but that at
least 10 ft (3.05 m) of good rock existed.
Cavity locations had a direct impact on the location of shaft size step-downs, cased or
uncased shaft design and final tip elevations. The criteria used to offset the effect of
cavities was to allow no shaft tip to be founded with less than 5 ft (1.52 m) of rock
Fig. 11.9 Plan view of MOTSU

Railroad in the vicinity of Boiling
Springs Dam (after Erwin & Brown,
1988).
above an underlying cavity and no cavity larger than 6 in (0.15 m) in height within the
next 5 ft (1.52 m). This was to prevent a punching-type failure from end bearing loads.
Cavities that were found required stepping-down to next smaller diameter of shaft and
casing 1 ft below the cavity (Fig. 11.10). The significant variability in the top rock
elevations also greatly affected shaft lengths.
Casings were installed and seated 1 ft (0.30 m) into rock using an oscillator prior to
excavation of the overburden. A large diameter borehole drill using a bucket auger was
used to clean out the overburden in the casings. A large diameter carbide-tipped spiral
auger was used to excavate the rock. Final cleanout of the shaft hole was by airlift.
Sinkholes developed in only two areas due to cavities encountered during drilling of the
shaft holes. These areas were simply backfilled and compacted and shaft installation
continued with no further problems. Three sinkholes were triggered during the drilling of
the contract borings. Six to eight sinkholes developed in the work area due to the
construction activities. Credit for the small number of sinkholes that developed is
probably due to that all holes in the overburden were cased at all times and that casings
were installed in the dry without use of drilling fluid or water.
Totally 124 drilled shafts were successfully installed at a cost of $2,000,000. The dry
technique of casing installation used by the contractor worked extremely well and helped
avoid potential installation problems that likely would have resulted if other techniques
had been used.
Fig. 11.10 Typical drilled shaft and

design assumptions for MOTSU
Railroad (after Erwin & Brown, 1988).
12
Loading test of drilled shafts in rock
12.1 INTRODUCTION
Drilled shafts cannot be readily inspected once they are constructed. On the other hand,
the performance of drilled shafts is highly dependent on the local geology and on the
construction procedure followed by the drilled shaft contractor. Hence it is not easy for
engineers to be assured that the constructed shafts comply with the design specifications.
Loading tests of drilled shafts are therefore highly desirable when it is feasible to perform
them.
Loading tests of drilled shafts are conducted for two general purposes:
1. to check the integrity of the test shaft and to prove that it is capable of sustaining the
applied loading as a structural unit;
2. to gain detailed information on load bearing and deformation characteristics of the
soil/rock and shaft system.
In the first instance the drilled shaft is constructed in the same manner as the production
shafts. The test shaft should sustain a load that is customarily at least twice the working
load without excessive displacement. In the second instance, the test shaft is instrumented
and usually loaded to failure by an appropriate definition. The instrumentation allows the
measurement of load and displacement along the length of the shaft. Such data allows
analyses to be made to obtain information on soil or rock resistance as a function of the
shaft displacement as well as the structural performance of the drilled shaft itself.
Loading tests of drilled shafts are expensive, and the cost should be carefully weighed
against the reduction in risk and assurance of satisfactory behavior that the loading test
provides. The extent of the test program depends on the availability of experience in
designing and constructing drilled shafts in a particular geological environment and the
capital cost of the works. A loading test of drilled shafts is most cost-effective when one
or more of the following conditions are present:
• Many drilled shafts are to be constructed, so even small savings on each shaft will
significantly reduce the overall construction cost.
• The soil/rock conditions are erratic or unusual.
• The structure is especially important or especially sensitive to displacements.
• The engineer has little or no experience in the project area.
Nearly all large drilled shaft projects should include at least one full-scale loading test.
However, it is not practical to test every shaft, even for the largest and most important
projects. Therefore, we can only test representative drilled shafts and extrapolate the
results to other shafts at the site. Table 12.1 lists the guidelines suggested by Engel
(1988) for determining the required number of pile load tests for typical projects.
Loading test of drilled shafts in rock 373
Table 12.1 Guidelines for determining the required

number of pile load tests (Engel, 1988).
Summation of length of all piles at the site
(ft) (m) Required number of load tests
0–6,000 0–1,800 0
6,000–10,000 1,800–3,000 1
10,000–20,000 3,000–6,000 2
20,000–30,000 6,000–9,000 3
30,000–40,000 9,000–12,000 4
12.2 AXIAL COMPRESSIVE LOADING TEST
12.2.1 General comments

The objectives of an axial compressive loading test are usually
1. to determine the ultimate bearing capacity of the shafts, relating this to the design
parameters;
2. to separate the total resistance contributed by the side shear and end bearing capacity;
3. to determine the stiffness of the soil/rock and shaft system at design load. A back
analysis of the tests data will enable the soil/rock modulus to be evaluated, and hence
the deformation of shaft groups may be predicted with greatly increased confidence.
These various objectives will necessitate a carefiilly chosen test procedure and
instrumentation program.
Considering the purposes of loading tests, the test shafts should be representative of
the production shafts. It is, therefore, critical that the test shafts be founded in the same
formation(s) as the production shafts and the construction procedures that are expected to
be used with the production shafts also be used with the test shafts. Since the capacity of
a full-scale drilled shaft in rock is usually very large, engineers may be tempted to
determine the values of unit shaft and base resistance from tests on small-diameter shafts
to reduce the cost of loading tests. It is found, however, that the unit ultimate resistances
developed by a small-scale drilled shaft are much higher than those developed by a full-
sized drilled shaft (O’Neill et al., 1996). So the unit shaft and base resistances determined
from tests on shafts with diameters much smaller than those of the production shafts can
be unconcervative. Recent practice in the United States has been to use test shafts in rock
that have approximately the same diameter and depth as and are constructed in a manner
similar to the production shafts (O’Neill & Reese, 1999).
12.2.2 Methods of applying loads
(a) Conventional loading test

The following conventional methods can be used to apply the compressive load on the
test shaft:
1. A platform is constructed on the head of the shaft and a mass of heavy material, termed
“kentledge”, is placed on the platform.
2. A bridge, carried on temporary supports, is constructed over the test shaft and loaded
with enough dead weight. The ram of a hydraulic jack, placed on the shaft head, bears
on a cross-head beneath the bridge beams, so that a total reaction equal to the weight
of the bridge and its load may be obtained.
3. Reaction shafts capable of withstanding an upward force are constructed on each side
of the test shaft, with a beam tied down to the head of the reaction shafts and spanning
the test shaft. A hydraulic jack on the head of the test shaft applies the load and
obtains a reaction against the underside of the beam (Fig. 12.1).
4. Ground or rock anchors are constructed to transfer the reaction to stiff strata below the
level of the shaft base.
For methods 1 and 2, more or less any material available in sufficient quantity can be
used as the dead weight. Specially-cast concrete blocks or pigs of cast iron may be hired
and transported to the site. The cost of transport to and from the site is a significant
factor. Regular-shaped blocks have the advantage that they may be stacked securely, and
are unlikely to topple unexpectedly. Sheet steel piling, steel rail, bricks, or tanks full of
sand or water have been used as the dead weight from time to time. It is important that
the mass of material be stable at all times during and after the test.
The reaction beams for methods 2 and 3 are subjected to high bending and buckling
stresses and they should be designed to carry the maximum load safely. The maximum
safe load should be clearly marked on the beam so that it is not inadvertently exceeded
during a test.
Stresses are transferred from the supports (method 2), reaction shafts (method 3) or
ground anchors (method 4) through the soil/rock mass, and such stresses can influence
the behavior of the test shaft. Therefore, the supports, reaction shafts or ground anchors
must be located well away from the test shaft to minimize this effect. Whitaker (1975)
recommends that the supports be more than 1.25 m (4 ft) away from a test pile, to
minimize the effect of the supports on pile settlement. He also recommends that any
reaction pile should be at least three test-pile diameters from a test pile, center to center,
and in no case less than 1.5 m (5 ft). The specifications of the American Society for
Testing and Materials for piling, ASTM D-1243, require that a clear distance of at least
five times the maximum of the diameters of the reaction or test piles must exist between a
test pile and each reaction pile (ASTM, 1995). According to O’Neill and Reese (1999), a
3.5 diameter center-to-center spacing between each reaction shaft and the test shaft is
adequate to minimize the reaction shaft and test shaft interaction for loading tests of large
diameter drilled shafts in cohesive soils or rocks.
Fig. 12.1 Compressive loading test

using reaction shafts.
Because the upper portion of an anchor cable does not usually transfer load to the
soil/rock, ground anchors can be placed closer to the test shaft than can reaction shafts. If
reaction shafts are constructed in such way that their uplift capacity is only developed in
a stratum far below the base of the test shaft (the bond between the reaction shaft and the
soil/rock to a depth well below the base of the test shaft is eliminated), the reaction shafts
can be as close to the test shaft as feasible. However, the construction disturbance of the
soil/rock around the test shaft still should be avoided.
The main disadvantages of ground anchors as a reaction system are the axial flexibility
of the anchor tendons and the lack of lateral stability of the system. A multiple ground
anchor system is to be preferred and each anchor should be proof tested to 130% of the
maximum load before use. To reduce the extension of the cables during the test it is usual
to pre-stress the anchors to as high a proportion of the maximum load as possible. Hence
a suitable reaction frame and foundation must be provided to carry this load safely.
(b) Osterberg cell loading test

Osterberg cell, named after its inventor, Jorj Osterberg, was first used in an experimental
drilled shaft in 1984 (Osterberg, 1984). Because of its advantages over the conventional
test systems, Osterberg cell is now widely used in axial loading test of drilled shafts.
Figure 12.2 illustrates schematically the difference between a conventional load test
and an Osterberg cell load test. A conventional test loads the drilled shaft in compression
at its top using an overhead reaction system or dead weight. Side shear Qs and end
bearing Qb combine to resist the top load Q and the engineer can only separate
Fig. 12.2 Comparison of conventional

load test and Osterberg cell load test.
these components approximately by analysis of strain or compression measurements
together with modulus estimates.
An Osterberg cell test also loads the drilled shaft in compression, but from its bottom.
As the Osterberg cell expands, the end bearing Qb provides reaction for the side shear Qs,
and vice versa, until reaching the capacity of one of the two components or until the
Osterberg cell reaches its capacity. Tests using the Osterberg cell automatically separate
the end bearing and side shear components. When one of the components reaches
ultimate capacity at an Osterberg cell load Qb, the required conventional top load Q to
reach both side shear and end bearing capacity would have to exceed 2Qb. Thus, an
Osterberg cell test load placed at or near the bottom of a drilled shaft has twice the testing
effectiveness of that same load placed at the top.
Osterberg cells have very large pistons, which makes it possible to apply very large
loads with relatively small jack pressures (see Table 12.2). By placing two or more
Osterberg cells on the same plane, the test capacity can be significantly increased. On
January 30, 2001, Loadtest, Inc. of the USA conducted a load test of 151 MN (17,000
tons) on an 8 ft (2.44 m) diameter, 135.5 ft (41.3 m) deep drilled shaft in Tucson,
Arizona, by utilizing three 34 in (870 mm) diameter Osterberg cells on a single plane
located approximately 28.5 ft (8.69 m) above the base of the shaft (LOADTEST, 2001).
Figure 12.3 shows the typical arrangement for an Osterberg cell load test of a drilled
shaft. The load being applied to the drilled shaft is determined by recording the pressure
and converting it to force from a pre-determined calibration curve. The upward

movement of the bottom and top of the shaft, and the downward movement of the bottom
of the shaft are measured by telltales and/or gauges. These measurements can be used to
obtain the side resistance versus side movement curve and the base resistance versus base
movement curve, as illustrated in Figure 12.4 for a test of a rock-socketed shaft in
Apalachicola River, Florida. Osterberg (1998) presented a method for constructing the
load-displacement curve equivalent to applying the load at the top of the shaft from the
side resistance versus side movement curve and the base resistance versus base
movement curve, which will be discussed in detail in Section 12.2.5.
Table 12.2 Size and load capacity of Osterberg cells
(LOADTEST, 2001).
Nominal Diameter Nominal Capacity*
(in) (mm) (kips) (MN)
9 230 450 2.00
13 330 870 3.87
21 540 2,000 8.90
26 670 3,640 16.2
34 870 6,150 27.4
*
The Osterberg cell applies a bi-directional load. The total test capacity is twice the nominal
Osterberg cell capacity.
Fig. 12.3 Typical arrangement of

Osterberg cell load test (after Ernst,
1995).
Fig. 12.4 Osterberg cell test load-

displacement curves of a rock-socketed
shaft in Apalachicola River, Florida
(after Osterberg, 1998).
A loading test with an Osterberg cell at one location as shown in Figures 12.2 and 12.3 is
limited by the magnitude of the side shear or base bearing resistance, whichever is
smaller. It is possible, however, to place Osterberg cells at two or more locations within
the drilled shaft. The first innovative application of this technique was a test for the
Alabama Department of Transportation in 1994 (Goodwin et al., 1994; O’Neill et al.,
1997). Figure 12.5 is a schematic of that test. The arrangement of the two 26.7 MN
(3,000 ton) Osterberg cells was such that it was possible to measure the base bearing
resistance, the side shear resistance in the socket in the chalk formation, the side shear
resistance in the cased portion of the shaft above the chalk, and the side shear resistance
in the chalk upon reversal of the direction of load.
The Osterberg cell test method offers a number of potential advantages over the
conventional loading test methods (Schmertmann & Hayes, 1997):
1. Economy: The Osterberg cell test is usually less expensive to perform than a
conventional static load test despite sacrificing the Osterberg cell. Osterberg cell tests
are typically 1/3 to 3/2 the cost of conventional tests, with the comparative cost
reducing as the test load increases.
2. High load capacity: especially for rock sockets.
3. Separation of side shear and end bearing components: The Osterberg cell test
automatically separates the side shear and end bearing components. It also helps
determine if construction techniques have adversely affected each component.
4. Improved safety: The test energy lies deeply buried and there is no overhead load.
5. Reduced work area: The work area required to perform an Osterberg cell test is much
smaller that that required by a conventional load system.
6. Over-water or battered shafts: Although often difficult to test conventionally, testing
over water or on a batter poses no special problems for the Osterberg cell load test
method.
7. Static creep and setup (aging) effects: Because the Osterberg cell test is static and the
test load can held for any desired length of time, information about the creep behavior
of the side shear and end bearing components can be obtained. The aging effects at
any time after installation can also be measured conveniently.
The Osterberg cell test method also has some limitations compared to the conventional
loading test methods. These include (Schmertmann & Hayes, 1997):
1. Advance installation required: The Osterberg cell must be installed prior to
construction of the shaft.
2. Balanced component required: An Osterberg cell test usually reaches the ultimate load
in only one of the two resistance components. The test shaft capacity demonstrated by
the Osterberg cell test is limited to two times the capacity of the component reaching
ultimate. Also, once installed the Osterberg cell capacity cannot be increased if
inadequate. To use the Osterberg cell efficiently the engineer should first analyze the
expected side shear and end bearing components and either attempt to balance the two
to get the most information from both or unbalance them to ensure the preferred
component reaches ultimate first. The introduction of multi-level Osterberg cell test as
discussed earlier can mitigates this limitation, allowing the engineer to obtain both
ultimate end bearing and ultimate side shear values in cases where the end bearing is
less than the side shear.
3. Equivalent top load curve: Although the equivalent static top load-displacement curve
can be estimated with conservatism, it remains an estimate.
4. Sacrificial Osterberg cell: The Osterberg cell is normally considered expendable and
not recovered after the test is completed. However, grouting the cell after completion
of the test allows using the tested shaft as a load carrying part of the foundation.
Fig. 12.5 Schematic of two level

Osterberg cell test in Alabama (after
O’Neill et al., 1997).
(c) Statnamic loading test

Statnamic load test was developed jointly in Canada and the Netherlands in the early
1990s (Bermingham & Janes, 1989; Middendorp et al., 1992). The principle of the
Statmanic load test is shown in Figure 12.6(a). Reaction mass is placed on the top of the
shaft. Beneath the reaction mass is a small volume of propellant (fast-expanding solid
fuel) and a load cell. By burning the propellant and propelling the reaction mass upward
off the shaft at accelerations up to 20 g, a load is generated. Since the mass is in contact
with the shaft prior to the test, the force associated with propelling of this mass acts
equally and oppositely onto the shaft. The reaction mass, usually rings of concrete or
steel, needs to be only 5% of the total load to be applied to the shaft. For reasons of
safety, the reaction mass is contained within a metal sheath that is also filled with an
energy absorber, such as dry gravel, that will cushion the impact of the mass as it fall
back upon the head of the shaft.
During a test, a high-speed data acquisition system scans and records the load cell,
displacement transducers, accelerometers and embedded strain gauges. The test
measurements provide a high degree of resolution fully defining the shaft load and
displacement response with up to 100,000 data points recorded during a typical ½ second
test. Because the measured Statnamic force includes some dynamic forces, some
interpretation of the data is necessary, as illustrated in Figure 12.6(b). Since the duration
of the axial Statnamic test is adequately longer than the natural period of the drilled shaft,
the entire drilled shaft remains in compression and a simple model can be used to
determine the static load acting on the shaft as follows (AFT, 2002):
Qstatic=QSTN−Ws(as/g)−cνs
(12.1)
where Qstatic is the derived static load acting on the shaft; QSTN is the measured Statnamic
force; Ws(as/g) is the inertia force; cνs is the damping force; Ws is the weight of the drilled
shaft; c is the damping coefficient of the drilled shaft and soil/rock system; and as and νs
are respectively the measured acceleration and velocity of the drilled shaft. c is the only
unknown in Equation (12.1) and can be determined using the principles of the Unloading
Point Method (UPM) (Middendorp et al., 1992; Brown, 1994).
Statnamic load test appears to offer a number of advantages over other types of tests,
including (AFT, 2002):
Fig. 12.6 (a) Sketch of Statnamic

loading test system; and (b) Statnamic
load-displacement curve and
interpreted static load-displacement
curve.
1. Statnamic load test is quick and easily mobilized.
2. Shaft performance is measured cost-effectively.

3. Statnamic load test is repeatable and the test shaft re-useable.
4. Statnamic load test does not require any reaction system.
5. Statnamic load test requires no special construction procedures so installation is more
representative of the actual production shaft construction. This allows Statnamic load
test to be used for random quality control testing and for problematic shafts.
6. The system is flexible and adaptable, e.g., single shafts or shaft groups can be tested
for compression loading and also lateral loading characteristics.
7. The test is quasi-static, and does not produce harmful compression and tension
stresses, which have the potential of damaging the shaft.
Statnamic load test also has limitations including (AFT, 2002):
1. Rate of loading precludes long term displacements.
2. Currently maximum Statnamic test load is 32 MN (3,600 tons).
3. Currently Statnamic load test method cannot be used for uplift tests.
12.2.3 Instrumentation
(a) Measurement of load

Calibrated pressure gauges are sometimes used in conventional load tests to determine
the load applied to the test shaft by measuring the hydraulic pressure in the loading ram.
However, calibrated pressure gauges may only provide acceptable accuracy of load
measurement for increasing load because the unloading cycle is usually nonlinear owing
to friction within the hydraulic jack. Therefore, load measurement is preferably carried
out using a load cell. The following lists four commonly adopted loadmeasuring devices
(Fleming et al., 1992):
1. hydraulic load cell capsules, maximum capacity 4.0 MN (450 tons) and accuracy
about±1%
2. load columns, maximum capacity 8.9 MN (1,000 tons) and accuracy about±1%
3. proving rings, maximum capacity 1.8 MN (200 tons) and accuracy about±0.5%
4. strain gauged load cells of various types, maximum capacity 1.8 MN (200 tons) and
accuracy about±0.5%
For very large loads, an array of load columns or strain gauged load cells may be used. It
is important that the load cell is calibrated regularly and that, for those devices sensitive
to temperature, suitable corrections are made.
When tests are run to obtain information on load transfer in side shear and end bearing
resistance, several methods can be used to measure the distribution of load along the
length of the drilled shaft, some of which being briefly described below (Fleming et al.,
1992; O’Neill & Reese, 1999):
Sister bars. The load distribution along the length of a drilled shaft can be obtained using
sister bars. A sister bar is a section of reinforcing steel at the middle of which is placed a
strain transducer. The sister bar is tied to the rebar cage and its leadwires routed to the
surface. The strain transducer can be a vibrating wire gauge or an electrical foil resistance
strain gauge. The vibrating wire transducer has the advantage that it tends to be stable
over longer periods of time than an electrical resistance transducer because the later is
quite sensitive to the invasion of moisture. However, electrical resistance transducers are
more adaptable to data acquisition systems than vibrating wire transducers.
Sister bars of both types are currently the most popular instruments for measuring load
distribution along the length of drilled shafts. The electrical output can be converted to
strain in the steel rebar through an appropriate calibration factor, which can then be
assumed to be equal to the strain in the concrete section. The internal load in the shaft can
then be obtained by multiplying the axial stiffness of the test shaft by the strain obtained
at the depth of interest.
It is important to place sister bars at opposite ends of diagonals at any level so that the
averaged readings cancel any bending effects that may occur. It is recommended that two
or four gauges be placed at each level at which load is to be measured.
Mustran cells. The Mustran cell is mounted on the rebar cage before inserting it into the
borehole, in a manner similar to that for sister bars. Because of its electrical circuitry, the
Mustran cell indicates strains that are larger than, but proportional to, the actual strains in
the concrete. This feature is advantageous for testing large-diameter drilled shafts
subjected to relatively small loads. Data from Mustran cells are collected and interpreted
in a manner very similar to that for sister bars. As with sister bars, it is good practice to
place Mustran cells at opposite ends of diagonals at any level so that the averaged
readings cancel any bending effects that may occur.
Telltales. Telltales are unstrained metal rods that are inserted into one or more tubes that
prevent them from bonding to the concrete in the shaft. The telltales extend to a series of
depths along the length of the shaft. The shortening of the test shaft over a particular
length can be found by using displacement transducers to measure the difference in the
movement of the shaft head and the top of the unstrained rod. Such measurements must
be made for each of the telltales and for each of the applied loads.
Again, to cancel any unintended bending effects, it is important to install telltales in
pairs, at opposite ends of diagonals at each depth. For a particular applied load, the
average deformation measured at each depth can be used to plot a compression of the
shaft versus depth curve. Differentiation of this deformation versus depth curve with
respect to depth will yield the strain of the shaft as a function of depth. The internal load
in the shaft can then be obtained by multiplying the axial stiffness of the test shaft by the
strain obtained at the depth of interest.
Pressure cells. Pressure cells placed at the bottom of the cage can be used to measure the
base resistance directly. Pressure cells are accurate because they do not require the
assumption of a value for the Young’s modulus of the concrete. Since stresses across
drilled shaft bases are generally uniform, the base resistance can be simply obtained by
multiplying the average pressure from the cells with the contact area of the shaft base.
(b) Measurement of displacement

The most commonly used systems to measure the head displacement of the test shaft are
dial gauges or electronic displacement transducers such as LVDT’s (linear variable
differential transformers) that are held by stable reference beams. Dial gauges are simple,
robust, mechanical precision instruments, and as such they should be carefully stored and
maintained. The reference beams should be long enough so that they can be supported by
firm foundations well away from the test shaft and reaction shafts. If possible, adjacent
drilled shafts at least four shaft diameters from the test shaft and reaction shafts can be
used as the support foundations. It is noted that a compromise must frequently made
between a long beam prone to vibration and temperature-induced displacements, and a
shorter beam with support foundations in the zone of influence of the test shaft or
reaction shafts.
Optical leveling may also be used to measure the head displacement of the test shaft.
The accuracy of an optical leveling system may be poorer than that of dial gauges by a
factor of at least 10. However the absolute accuracy of the system may be of a similar
order to that of dial gauges, particularly in situations where it is difficult to establish a
stable reference beam. With an optical system it is easy to arrange for the instrument and
reference point to be well away the zone of disturbance (Fleming et al., 1992).
When tests are run to obtain information on load transfer in side shear and end bearing
resistance, telltales as described earlier can be used to measure the displacement of the
test shaft from point to point along its length.
12.2.4 Test procedures

The procedures for conducting conventional loading test are given in the ASTM Test
Designation D-1243 (ASTM, 1995). The U.S. Corps of Engineers Manual on the “Design
of Pile Foundations” recommends the following three methods of loading (ASCE, 1993):
1. Slow, maintained load test method. This is the most common test procedure and is
referred to as “standard loading procedure” in the ASTM Test Designation D-1243. In
this method, the shaft is loaded in eight equal increments up to a maximum load,
usually twice a predetermined allowable load. Each of the eight load increments is
placed on the shaft very rapidly (as fast as the pump can raise the load, which usually
takes about 20 seconds to 2 minutes) and maintained until zero movement is reached,
defined as 0.25 mm/h (0.01 in/h). The final load, the 200 percent load, is maintained
for a duration of 24 hours. This procedure is very time consuming, requiring from 30
to 70 or even more hours to complete. It should be noted that the phrase “zero
movement” is very misleading: the “zero” movement rate is equal to a movement of
more than 2 m (7 ft) per year.
2. Constant rate of penetration (CRP) test method. In a CRP test, the load is applied to
cause the shaft head to settle at a predetermined constant rate, usually in the vicinity of
0.25 mm (0.01 in) per minute to 2.5 mm (0.1 in) per minute, depending on the
soil/rock type. The duration of the test is usually 1 to 4 hours, depending on the
variation used. The particular advantage of the CRP test is that it can be conducted in
less than one working day. A disadvantage is that ordinary pumps with pressure
holding devices like those used for “slow” tests are difficult to use for the CRP test. A
more suitable pump is one that can provide a constant, nonpulsing flow of oil.
3. Quick maintained load test method. In this test, the load is applied in increments of
about 10 percent of the proposed design load and maintained for a constant time
interval, usually about 2 to 15 minutes. The duration of this test will generally be
about 45 minutes to 2 hours, again depending on the variation selected. The advantage
of this test, like the CRP test, is that it can be completed in less than one working day.
Unlike the CRP test, however, no special loading equipment is required.
For particular projects special loading paths may be called for, to simulate repeated
loading for example. Such loading paths my easily be arranged but careful supervision is
necessary. Short-period, cyclic or sinusoidal loading requires the use of sophisticated
servo-controlled equipment.
The loading procedure for the conventional load test method can be easily applied to
the Osterberg cell load test method.
It is essential to record all the relevant data throughout the test, including the load,
displacement, time, problems, and unexpected occurrences.
12.2.5 Interpretation of test data

A considerable amount of data may be obtained from an axial loading test, and with more
sophisticated instrumentation a greater understanding of the soil/rock and shaft
interaction may be achieved. Interpretation of the axial loading test data may be carried
out on several levels as described below.
(a) Load-displacement curve of shaft head

From a conventional load test, the load-displacement curve of the shaft head can be
directly obtained. For an Osterberg cell load test, the side resistance versus side
movement curve and the base resistance versus base movement curve can be used to
construct the load-displacement curve equivalent to applying the load at the top of the
shaft (Osterberg, 1998). This is done by determining the side resistance at an arbitrary
displacement point on the side resistance versus side movement curve. If the shaft is
assumed rigid, the top and bottom of the shaft move the same amount and have the same
displacement but different loads. By adding the side resistance and the base resistance at
the same displacement, a single point on the top equivalent loaddisplacement curve is
obtained. By repeating this process for different displacement points, the top equivalent
load-displacement curve can be obtained (see Fig. 12.7). It is very often that the side
shear or the end bearing reaches ultimate before the other (In Figure 12.7, the side shear
reaches ultimate before the end bearing). In this case, two procedures are possible. The
first, which is extremely conservative, is to assume that the component that has not
reached ultimate has also reached ultimate and that no further load increase occurs as
displacement increases. The other more likely procedure, as shown in Figure 12.7, is to
extrapolate the curve that has not reached ultimate. The procedure for constructing the
top equivalent load-displacement curve is based on the following assumptions
(Osterberg, 1998):
1. The side shear-displacement curve for upward movement of the shaft is the same as the
downward side shear-displacement component of a conventional load test.
2. The base resistance-displacement curve obtained from an Osterberg cell load test is the
same as the base resistance-displacement component of a conventional load test.
3. The shaft is considered rigid. Typically, the compression of the shaft at ultimate loadis
1–3 mm.
For a Statnamic test, the Statnamic curve can be corrected to obtain the equivalent static
load-displacement curve as described in Section 12.2.2.
(b) Estimation of ultimate load

The top load-displacement curve is commonly used to determine the ultimate load
capacity of the test shaft. A number of criteria for defining the ultimate load capacity
have been developed and discussed by different researchers, some of which are listed
below (Tomlinson, 1977; Fellenius, 1980; ASCE, 1993):
1. The load at which displacement continues to increase without further increase in load.
2. The load causing a certain amount of total displacement, such as 1 in (25 mm), 10% of
the base diameter of the shaft, elastic displacement plus 1/30 of the shaft diameter, or
elastic compression of the shaft plus 0.15 in (4 mm) plus 1/120 of the shaft diameter.
3. The load causing a certain amount of plastic displacement, such as 0.25 in (6 mm).
4. The load at a defined plastic to elastic displacement ratio, such as 1.5.
5. The load at a defined slope of the load-displacement curve, such as 0.01 in (0.25 mm)
per 1 ton (10 kN).
6. Load-displacement curve interpretation
a. Maximum curvature—plot log total displacement versus log load; the ultimate load
capacity is at the point of maximum curvature.
b. Tangents—plot tangent lines to the initial and failure portions of the
loaddisplacement curve; the ultimate load capacity is at the intersection of the two
tangent lines.
c. Inverse slope—divide each load value by its corresponding displacement values and
plot the resulting value against the displacement; the plotted values fall on a
straight line and the inverse slope of the line is the ultimate load capacity.
Different criteria may result in very different values of the ultimate load capacity. It is
important to choose the appropriate criterion/a and check the obtained value(s) of the
ultimate load capacity. The method often used by the U.S. Corps of Engineers and
presented below is a good example on how to determine the ultimate load capacity:
The following method has often been used by the U.S. Corps of Engineers
and has merit: determine the load that causes a plastic displacement of
0.25 in (6 mm); determine the load that corresponds to the point at which
the load-displacement curve has a significant change in slope; and
determine the load that corresponds to the point on the load-displacement
curve that has a slope of 0.01 in per ton (0.25 mm per 10 kN). The
average of the three loads determined in this manner would be considered
the ultimate axial capacity of the pile. If one of these three procedures
yields a value that differs significantly from the other two, judgment
should be used before including or excluding this value from the average.
A suitable factor of safety should be applied to the resulting axial pile
capacity.
Fig. 12.7 Construction of equivalent

top load-displacement curve (b) from
Osterberg cell load test curves (a)
(after Osterberg, 1998).
(c) Distribution of load along depth

With the instruments described in Section 12.2.3, the distribution of load along the
length of the shaft can be obtained as shown in Figure 12.8. The load distribution curves
clearly show the contribution of the side shear and the end bearing. The shape of the load
distribution curves also reflects the distribution of the side shear resistance along the
length of the shaft.
(d) Load transfer (t-z or τ-w and q-w) curves

It is also possible to obtain the load transfer (t-z or τ-w and q-w) curves for
instrumented shafts. For example, to obtain the τ-w curve at depth zi, the following
procedure can be followed (see Fig. 12.9):
1. For a test load, determine the movement of the shaft relative to the soil or rock at depth
zi: The relative movement at depth zi is obtained by subtracting the shaft shortening
over the distance zi from the measured displacement at the top of the shaft. The shaft
shortening over the distance zi is simply obtained by dividing the cross-hatched area
[see Fig. 12.9(a)] by the shaft axial stiffness which is the product of the cross-sectional
area and the composite Young’s modulus of the steel and concrete.
2. For a test load, determine the side shear resistance at depth zi: The unit side shear
resistance at depth zi is obtained by dividing the slope of the load distribution curve at
depth zi [see Fig. 12.9(a)] by the perimeter length of the shaft.
3. Plot the movement of the shaft and the side shear resistance determined in Steps 1 and
2 in the τ versus w plot [see Fig. 12.9(b)].
4. Repeat Steps 1 to 3 for all other test loads and draw the τ-w curve at depth zi by
connecting the points obtained.
At other depths, the τ-w curves can be obtained by simply repeating the above steps.
For the q-w curve, the base resistance at a test load can be determined by simply
taking the base load from the load distribution curve. The base movement of the shaft at a
test load can be determined following the method in Step 1 above.
Fig. 12.8 Typical load distribution

curves.
Fig. 12.9 Construction of load transfer

curves.
12.3 AXIAL UPLIFT LOADING TEST
Axial uplift loading tests are similar to axial compressive loading tests, and maintained
load or constant rate of uplift procedures may be used. Figure 12.10 shows a loading
system for uplift loading tests. To avoid bending the test shaft, two jacking points on
either side of the test shaft are used and care is required to load the jacks evenly.
Fig. 12.10 Uplift loading test using

ground as reaction.
Uplift test of shafts may be carried out using a constant rate of uplift procedure which is
similar to the constant rate of penetration (CRP) test. However, since the full uplift
capacity of a drilled shaft is normally mobilized at a displacement smaller than that
required for the full compressive capacity, the imposed rate of uplift should be reduced,
say to 0.1 to 0.3 mm per minute.
Interpretation of the uplift load test results can follow the same procedures as for the
axial compressive load tests. Similarly, load-displacement curves, ultimate uplift load
capacity, load distribution along the depth, and load transfer curves (only t-z or τ-w
curves for uplift loading tests) can be obtained.
12.4 LATERAL LOADING TEST
12.4.1 General comments

Drilled shafts are frequently used to carry lateral loads because of their high moment of
inertia. The main purpose of lateral load tests is to verify the behavior of production
shafts or to provide data for constructing the load transfer (p-y) curves. The basis for
conducting a lateral load test can be the ASTM Standard D 3966 (1995) or Eurocode 7
(1994) modified to satisfy the specific project requirements.
12.4.2 Methods of applying loads
(a) Conventional loading test

A lateral load test is most easily conducted by pulling a pair of shafts together or jacking
them apart (Fig. 12.11). In this manner, two lateral load tests can be conducted
simultaneously. It is important to note that, however, spacing between the two shafts
should be such that the shaft-to-shaft interaction is minimized. Because of the difficulty
of applying load exactly at the ground line, load is normally applied at a point above the
ground line, resulting in both shear and moment at the ground line. The shaft head is
generally unrestrained, giving a free-head shaft boundary condition and making the data
easy to analyze. The load is applied with a hydraulic jack and measured by a load cell
adjacent to the jack in a manner similar to that for axial loading tests. A spherical bearing
head should be used to minimize eccentric loading.
Fig. 12.11 Lateral loading test of a pair

of shafts.
(b) Osterberg cell loading test

Osterberg cells have been used to test rock sockets to duplicate the behavior of laterally
loaded drilled shafts in rock (O’Neill et al., 1997). This is done by inserting an Osterberg
cell vertically into the socket, casting concrete around the cell and using the cell to jack
the two halves of the socket apart (Fig. 12.12). The lateral load applied to the rock per
unit socket length is computed by dividing the load in the cell by the length of the socket.
Lateral displacement is measured by using LVDT’s that connect between the plates. It
need be noted that, however, the stress and strain conditions in the rock for the lateral
Osterberg cell test, in which the halves of the socket are jacked apart, are different from
those for a drilled shaft that translates laterally within the rock without splitting.
(c) Statnamic loading test

Statnamic devices can also be used to test drilled shafts laterally by assembling them
horizontally on a special “sled” (Fig. 12.13). Lateral Statnamic loading test has the
advantages of closely modeling lateral loading events, such as wind, seismic load, ship
impact and wave action, and of no requirement for the construction of a costly reaction
shaft. As for the axial Statnamic loading test, however, analyses need be conducted to
consider the effects of shaft and soil/rock inertia and loading rate.
The load measurement devices for axial loading tests can be used for the measurement of
load in lateral loading tests. If the test is to derive the p-y curves, strain gauges may be
installed along the shaft to measure the bending moment. Differentiations of the moment
distribution along the depth can yield soil/rock resistance along the depth (see Section
12.4.5 for details).
Fig. 12.12 Lateral Osterberg cell test of

rock socket (Modified from O’Neill et
al., 1997).
Fig. 12.13 Lateral Statnamic loading

test of drilled shaft (from AFT, 2002).
12.4.3 Instrumentation
(a) Measurement of load
(b) Measurement of displacement

For lateral loading tests, the lateral ground line deflection and the shaft head rotation
should, as a minimum, be measured. The ground line deflection can be measured by
means of displacement instruments suspended from reference beams, similar to the way
settlement or uplift is measured in axial loading tests. Rotation can be measured by
measuring the lateral deflections at two points above the shaft head (Fig. 12.14) or the
vertical displacements at two points on the shaft head (Fig. 12.15). The shaft head
rotation is simply the difference in lateral deflections or vertical displacements at these
two points divided by the distance between them. A tiltmeter (Fig. 12.16) or an
inclinometer (Fig. 12.17) can also be used for measuring the shaft head rotation. The
inclinometer can measure the lateral deflection and rotation along the length of the shaft.
Strain gauges installed along the shaft can be used to measure the bending moment
distribution along the shaft. Integrations of the moment distribution along the depth can
yield lateral deflection along the depth (see Section 12.4.5 for details).
12.4.4 Test procedures

The procedures for conducting lateral loading tests are given in the ASTM Standard D
3966 (1995) and Eurocode 7 (1994). Like the axial loading tests, two principles should
guide the testing procedure: (1) The loading (static, repeated with or without load
reversal, sustained, or dynamic) should be consistent with that expected for the
production shafts; and (2) The testing arrangement should allow deflection, rotation,
bending moment, and shear at the ground line or at the point of load application to be
measured or computed.
12.4.5 Interpretation of test data

A lateral loading test may produce a considerable amount of data corresponding to the
instruments used. Interpretation of the lateral loading test data may be carried out on
several levels as described below.
(a) Load-deflection curve and load-rotation curve at ground line

From a conventional lateral loading test, the load-deflection curve and the load-rotation
curve at the ground line can be directly obtained. For a lateral Statnamic loading test, the
Statnamic curves can be corrected to obtain the equivalent static load-deflection curve
and the load-rotation curve at the ground line.
(b) Estimation of lateral load capacity

The lateral load-deflection curve can be used to estimate the lateral load capacity. For
example, the ultimate load can be taken as the load required to produce a specified
deflection at the ground line (e.g., 0.25 in).
(c) Distribution of deflection and bending moment along depth

Distribution of deflection along the depth can be directly obtained by means of electronic
inclinometers running up and down tracks in inclinometer tubing that has been cast in the
drilled shaft. The distribution of bending moment along the depth can be obtained from
the measurements of strain gauges installed along the shafts. It is important to consider
the changes of cross-section and modulus of the shaft along the depth when determining
the bending moments from the measurements of strains. Figure 12.18 shows the
distribution of bending moment from the lateral loading test of a drilled shaft.
There are several ways to determine the load transfer (p-y) curves from the test data,
depending the instruments used in the test.
Fig. 12.14 Measurement of shaft head

rotation by measuring lateral
deflections at two points above the
shaft head.

rotation by measuring vertical
displacements at two points on the
shaft head.
(d) Load transfer (p-y) curves

rotation using tiltmeter.
Fig. 12.17 Measurement of shaft

deflection and rotation using
inclinometer.
Fig. 12.18 Distribution of bending

moment along depth from lateral
loading test of a drilled shaft.
Using the load-deflection curve and the load-rotation curve at the ground line, computer
programs can be executed by varying the form of the p-y curves along the shaft until a
match is achieved in both deflection and rotation under all of the loads applied. More
accurate definition of p-y curves can be achieved if the distribution of deflection along
the depth is available.
The bending moment curves can be used to obtain the p-y curves by integrating and
differentiating the curves along the length of the shaft. The deflection can be obtained
with considerable accuracy by two integrations of the bending moment curves:
(12.2)
where y is the lateral deflection; Mis the bending moment; EpIp is the flexural rigidity of
the shaft; and z is the depth from the ground line. To determine the deflection accurately,
the deflection and slope at the ground line have to be measured accurately and it is
helpful if the shaft is long enough so that there are at least two zero-deflection points
along the shaft.
The resistance of the soil/rock can be obtained by two differentiations of the bending
moment curves:
(12.3)
where p is the resistance of the soil/rock. Differentiations can be performed numerically

along the shaft by using the data without adjustment (Matlock, 1970) or by fitting
analytical curves through the points of measured bending moment and then performing
differentiations mathematically. It need be noted that slight errors in the measured values
of the bending moment will cause large and unacceptable errors in the values of P.
12.5 INTEGRITY TEST OF DRILLED SHAFTS
The presence of any defects in a drilled shaft, such as cracks, waists or voids, may cause
a serious decrease of load capacity and an increase of displacement. It is, therefore,
important to conduct integrity test to detect the defects in drilled shafts. Many forms of
integrity test have been developed to detect flaws in deep foundations (Fleming et al.,
1992), two of which are frequently used for drilled shafts: the low strain integrity test
using the Pile Integrity Tester (PIT) and the cross-hole sonic logging test.
The low strain integrity test utilizes one-dimensional wave propagation. A small hand
held hammer impacts the top of the shaft, and an accelerometer attached to the top of the
shaft measures the impact and resulting reflections (Fig. 12.19). The measurement is then
analyzed for relevant reflections from the shaft toe or major shaft anomalies. A record
that shows a clear reflection from the shaft toe and no major reflections from intermediate
points indicates a sound shaft. Generally, shafts that contain anomalies show a significant
wave reflection from a shorter length and no toe reflection.
The low strain integrity test using the PIT can be applied to practically every shaft on
site due to its low cost and minimal shaft preparation. It is often the first alternative when
questions of shaft acceptability arise after the installation is completed. The low
Fig. 12.19 Low strain integrity test

using Pile Integrity Tester (PIT).
Fig. 12.20 Cross-hole sonic logging

test.
strain integrity test using the PIT is useful for selecting shafts for further testing. If it
determines obviously good or obviously bad shafts, the solution is clear. For tests
indicating marginal conditions further testing of another type may be desired. The low
strain integrity test using the PIT is described in detail in ASTM standard D5882.
The cross-hole sonic logging test measures sound velocity along the shaft between an
emitting sensor and a receiving sensor lowered down two tubes (Fig. 12.20). If the
measured velocity decreases rapidly from the sound velocity of homogeneous concrete
(about 4000 m/s), defects such as soil inclusion, cracks or segregation may exist. The
cross-hole sonic logging test requires that tubes be installed in the shaft prior to
concreting.
Table 12.3 lists the advantages and disadvantages of the two integrity test methods.
Table 12.3 Advantages and disadvantages of low
strain integrity test using PIT and cross-hole sonic
logging test.
Test Method Advantage Disadvantage
Low strain • No special preparation • Test interpretation limited if toe cannot be
integrity test needed seen due to excess length or multiple
using PIT section changes
• Quick, simple and
inexpensive
• Yields information on major
variations of quality or size
Cross-hole sonic • Works on drilled shafts of • Inspection tubes installed during shaft
logging test unlimited size or length construction
• Clear identification of • Tube debonding sometimes prevents wave
defects even at great depth transmission
• Need to wait for concrete hardening
References
AASHTO (1989). Standard Specifications for Highway Bridges. (14th ed.). American Association
of State Highway and Transportation Officials, Washington DC.
AASHTO (1994). LRFD Bridge Design Specifications. (1st ed.). American Association of State
Highway and Transportation Officials, Washington DC.
Abad, J., Caleda, B., Chacon, E., Gutierez, V., & Hidlgo, E. (1984). Application of geomechanical
classification to predict the convergence of coal mine galleries and to design their supports. 5th
Int. Cong. on Rock Mech., E, 15–19.
ABAQUS (1998). Finite Element Program, Version 5.8. Hibbitt, Karlsson and Sornsen, Inc.,
Providence, RI.
ACI (1995). Building Code Requirements for Structural Concrete (ACI318-95) and Commentary
(ACI318R-95). American Concrete Institute, Farmington Hills, MI.
AFT (2002). Statnamic Load Testing Overview. Applied Foundation Testing, Inc.
Alfonsi, P., Durville, J.-L., & Rachez, X. (1998). Quelques applications de la methode des elements
distincts en mecanique des roches. Bulletin des Laboratories des Ponts et Chaussees, 214, 31–
43.
Alfonsi, P., Durville, J.-L., & Rachez, X. (1999). Numerical modeling of a foundation on rock
slope using the distinct element method. Proc. 9th Int. Cong. on Rock Mech., 1, 71–76.
Amadei, B. (1983). Rock Anisotropy and the Theory of Stress Measurements, Lecture Notes in
Engineering. Eds: C.A.Brebbia & S.A.Orszag, Springer-Verlag, Berlin.
Amadei, B. (1988). Strength of a regularly jointed mass under biaxial and axisymmetric loading
conditions. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 25, 3–13.
Amadei, B., & Savage, W.Z. (1989). Anisotropic nature of jointed rock mass strength. J. Geotech.
Engrg., ASCE, 115(3), 525–542.
Amadei, B., & Savage, W.Z. (1993). Effect of joints on rock mass strength and deformability.
Comprehensive Rock Engrg.—Principle, Practice and Projects. Ed: J.A.Hudson,Pergamon
Press, 1, 331–365.
Amir, J.M. (1986). Pilling in Rock. A.A.Balkema Publishers, Rotterdam, Netherlands.
Andersson, J., Shapiro, A.M., & Bear, J. (1984). A stochastic model of a fractured rock conditioned
by measured information. Water Resources Research, 20(1), 79–88.
API (1982). Recommended Practice for Planning, Designing, and Constructing Fixed Offshore
Platforms. (13th ed.). API RP2A, American Petroleum Institute, Washington, DC.
API (1991). Recommended Practice for Planning, Designing, and Constructing Fixed Offshore
Platforms. (19th ed.). API RP2A, American Petroleum Institute, Washington, DC.
ARGEMA (1992). Design Guides for Offshore Structures: Offshore Pile Design. Ed: P.L.Tirant,
Éditions Technip, Paris, France.
ASCE (1993). Design of Pile Foundations: Technical Engineering and Design Guides as Adapted
from the US Army Corps of Engineers. ASCE Press, New York, NY.
ASCE (1996). Rock Foundations: Technical Engineering and Design Guides as Adapted from the
US Army Corps of Engineers. No. 16, ASCE Press, New York, NY.
ASCE (1997). Standard Guidelines for the Design and Installation of Pile Foundations. ASCE 20–
96, ASCE Press, New York, NY.
ASCE (1998). Geophysical Exploration for Engineering and Environmental Investigations:
Technical Engineering and Design Guides as Adapted from the US Army Corps of Engineers.
No. 23, ASCE Press, Reston, VA.
References 405
ASTM (1971). Standard Method of Test for Unconfined Compressive Strength of Intact Rock
Specimens. ASTM Designation D 2938–71a, American Society for Testing and Materials,
Philadelphia, PA.
ASTM (1995). Annual Book of ASTM Standards, Vol. 4.08, Soil and Rock. American Society for
Testing and Materials, Philadelphia, PA.
Aurora, R.P., & Reese, L.C. (1977). Field test of drilled shafts in clay shale. Proc. 9th Int. Conf on
Soil Mech. and Found. Engrg., Tokyo, Japan, 371–376.
Baecher, G.N. (1980). Progressively censored sampling of rock joint traces. Mathematical
Geology, 12(1), 33–40.
Baecher, G.N., & Lanney, N.A. (1978). Trace length biases in joint surveys. Proc. 19th U.S. Symp.
on Rock Mech., 1, 56–65.
Baecher, G.N., Lanney, N.A., & Einstein, H.H. (1977). Statistical description of rock properties
and sampling. Proc. 18th U.S. Symp. on Rock Mech., 5C1-8.
Baguelin, F. (1982). Rules for the structural design of foundations based on the selfboring
pressuremeter test. Symp. on Pressuremeter and its Marine Application, IFP, Paris, 347–362.
Baguelin, F., Jezequel, J.F., & Shields, D.H. (1978). The Pressuremeter and Foundation
Engineering. Trans Tech Publications.
Baker, C.N., Jr. (1985). Comparison of caisson load tests on Chicago hardpan. Drilled Piers and
Caissons II. Ed: C.N. Jr. Baker, ASCE, Denver, Colorado, 99–113.
Baker, C.N., Jr. (1994). Current U.S. design and construction practice for drilled piers. Proc. Int.
Conf. on Design and Construction of Deep Found., Orlando, FL, 305–323.
Balmer, G. (1952). A general analytical solution for Mohr’s envelope. American Soci. Test. Mater.,
52, 1260–1271.
Bandis, S.C. (1990). Mechanical properties of rock joints. Proc. Int. Symp. on Rock Joints, Loen,
Norway, Eds: N.Barton & O.Stephanson, Balkema, Rotterdam, 125–140.
Bandis, S.C., Lumsden, A.C., & Barton, N.R. (1981). Experimental studies of scale effects on the
shear behavior of rock joints. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 18(1), 1–21.
Banerjee, P.K., & Davies, T.G. (1978). The behavior of axially and laterally loaded single piles
embedded in nonhomogeneous soils. Geotechnique, 28(3), 309–326.
Barry, A.J., & Nair, O.B. (1970). In Situ Tests of Bearing Capacity of Roof and Floor in Selected
Bituminous Coal Mines. A Progress Report—Longwall Mining, US Bureau of Mines, RI 7406.
Barton, C.M. (1977). Geotechnical analysis of rock structure and fabric in CSA Mine NSW.
Applied Geomechanics Technical Paper 24, Commonwealth Scientific and Industrial Research
Organization, Australia.
Barton, N. (1976). The shear strength of rock and rock joints. Int. J. Rock Mech. Min. Sci. &
Geomech. Abstr., 13, 255–279.
Barton, N. (1987). Predicting the Behavior of Underground Openings in Rock. Manuel Rocha
Memorial Lecture, Lisbon, Oslo, Norwegian Geotech. Inst.
Barton, N., & Bandis, S.C. (1982). Effects of block size on the shear behavior of jointed rock. Proc.
23rd U.S. Symp. on Rock Mech., Berkeley, 739–760.
Barton, N., & Bandis, S.C. (1990). Review of predictive capabilities of JRC-JCS model in
engineering practice. Proc. Int Symp. on Rock Joints. Loen, Norway, Eds: N.Barton & O.
Stephansson, Balkema, Rotterdam, 603–610.
Barton, N., & Choubey, V. (1977). The shear strength of rock joints in theory and practice. Rock
Mech., 10, 1–54.
Barton, N., Lien, R., & Lunde, J. (1974). Engineering classification of rock masses for the design
of tunnel support. Rock Mech., 6, 189–236.
Barton, N., Loset, F., Lien, R., & Lunde, J. (1980). Application of the Q-system in design
decisions. Subsurface Space, Ed: M.Bergman, 2, 553–561.
Baycan, S. (1996). Field Performance of Expansive Anchors and Piles in Rock. PhD Dissertation,
Department of Civil Engineering, Melbourne, Australia.
References 406
Bermingham, P., & Janes, M. (1989). An innovative approach to load testing of high capacity piles.
Proc. Int. Conf On Piling and Deep Found., 1, 409–413.
Bieniawski, Z.T. (1968). The compressive strength of hard rocks. Tydskrif vir Natuurwetenskappe,
8, 163–182.
Bieniawski, Z.T. (1968). The effect of specimen size on compressive strength of coal. Int. J. Rock
Mech Min. Sci. & Geomech. Abstr., 5, 321–335.
Bieniawski, Z.T. (1973). Engineering classification of jointed rock masses. Trans South African
Inst. Civil Engrs., 15, 335–344.
Bieniawski, Z.T. (1974). Geomechanics classification of rock masses and its application in
tunneling. Proc. 3rd Int. Cong. on Rock Mech., Denver, IIA, 27–32.
Bieniawski, Z.T. (1976). Rock mass classification in rock engineering. Proc. Symp. on Exploration
for Rock Engrg., Ed: Z.T.Bieniawski, Cape Town, Balkema, 1, 97–106.
Bieniawski, Z.T. (1978). Determining rock mass deformability: experience from case histories. Int
J. Rock Mech Min. Sci. & Geomech. Abstr., 15, 237–248.
Bieniawski, Z.T. (1984). Rock Mechanics Design in Mining and Tunneling. Balkema, Rotterdam/
Boston.
Bieniawski, Z.T. (1989). Engineering Rock Mass Classifications. John Wiley, Rotterdam
Bieniawski, Z.T., & Van Heerden, W.L. (1975). The significance of in situ tests on large rock
specimens. Int. J. Rock Mech Min. Sci. & Geomech Abstr., 12, 101–113.
Bishnoi, B.L. (1968). Bearing Capacity of a Closely Jointed Rock. PhD thesis, Georgia Institute of
Technology, Atlanta.
Blyth, G.G.H., & M.H.de Freitas (1974). A Geology for Engineers. (6th ed.). London, Edward
Arnold.
BOO (1990). Foundation design. Building (construction) regulations 1990—Part VI. Prac. Note for
Authorized Persons and Registered Struct. Engineers., No. 141, 1995 Revision, Buildings
Department, Building Ordinance Office, Hong Kong.
Borgard, D., & Matlock, H. (1980). Simplified Calculation of p-y Curves for Laterally Loaded Piles
in Sand. Earth Technology Corporation, Inc., Houston, TX.
Bowles, J.E. (1996). Foundation Analysis and Design. (5th ed.). McGraw Hill, New York.
Brady, B.H.G., & Brown, E.T. (1985). Rock Mechanics for Underground Mining. George Allen &
Unwin, London.
Briaud, J.-L. (1986). Pressuremeter and deep foundation design. The Pressuremeter and its Marine
Applications: 2nd Int. Symp., ASTM STP 950, 376–405.
Briaud, J.-L., & Smith, T.D. (1983). Using the pressuremeter curve to design laterally loaded piles.
Offshore Technology Conf., Houston, Paper 4501, 495–502.
Bridges, M.C. (1975). Presentation of fracture data for rock mechanics. Proc. 2nd Australia-New
Zealand Conf. on Geomech., Brisbane, 144–148.
Brinch Hansen, J. (1961). The ultimate resistance of rigid piles against transversal forces.
Geoteknisk Institut. Bull., No. 12, Copenhagen.
Broms, B.B. (1964a). Lateral resistance of piles in cohesive soils. J. Soil Mech. Found. Div.,
ASCE, 90(2), 27–63.
Broms, B.B. (1964b). Lateral resistance of piles in cohesionless soils. J. Soil Mech. Found. Div.,
ASCE, 90(3), 123–156.
Brown, D.A. (1990). Construction and design of drilled shafts in hard pinnacle limestones.
Transportation Research Record, ASTM, 17(4).
Brown, D.A. (1994). Evaluation of static capacity of deep foundations from Statnamic testing.
Geotech. Testing J., 1277, 148–152.
Brown, D.A., Morrison, C., & Reese, L.C. (1988). Lateral load behavior of pile groups in sand. J.
Geotech. Engrg., ASCE, 114(11), 1261–1276.
Brown, E.T. (1970a). Strength of models of rock with intermittent joints. J. Soil Mech. Found Div.,
ASCE, 96,1935–1949.
References 407
Brown, E.T. (1970b). Modes of failure in jointed rock masses. Proc. 2nd Int. Cong. on Rock Mech.,
Belgrade, 2, 3–42.
Brown, E.T. (1993). The nature and fundamentals of rock engineering. Comprehensive Rock
Engrg.—Principle, Practice and Projects. Ed: J.A.Hudson, Pergamon Press, 1, 1–23.
Butterfield, R, & Douglas, R.A. (1981). Flexibility Coefflcients for the Design of Piles and Pile
Groups. CIRIA Tech. Note 108, CIRIA, London.
Buttling, S. (1986). Testing and instrumentation of bored piles. Proc. 4th Nanyang Technological
Institute Geotechnical Seminar on Field Instrumentation and In-site Measurements, Singapore,
211–218.
Buttling, S., & Lam, T.S.K. (1988). Behavior of some instrumented rock socket piles. Proc. 5th
Australian-New Zealand Conf. on Geomech., Sydney, 526–532.
Call, R.D., Savely, J.P., & Nicholas, D.E. (1976). Estimation of joint set characteristics from
surface mapping data. Proc. 17th U.S. Symp. on Rock Mech., 2B2, 1–9.
Call, R.D., Savely, J.P., & Pakalnis, R. (1981). A simple core orientation technique. 3rd Int. Conf.
on Stability in Surface Mining. Vancouver, British Columbia, Canada.
Cameron-Clarke, I.S., & Budavari, S. (1981). Correlation of rock mass classification parameters
obtained from bore core and in-situ observations. Engineering Geology, 17, 19–53.
Carrubba, P. (1997). Skin friction of large-diameter piles socketed into rock. Can. Geotech. J.,
Ottawa, Canada, 34(2), 230–240.
Carter, J.P., Booker, J.R., & Yeung, S.K. (1986). Cavity expansion in cohesive frictional soils.
Geotechnique, 36(3), 349–358.
Carter, J.P., & Kulhawy, F.H. (1988). Analysis and Design of Drilled Shaft Foundations Socketed
into Rock. Report EL-5918, Electric Power Research Institute, Palo Alto, CA.
Carter, J.P., & Kulhawy, F.H. (1992). Analysis of laterally loaded shafts in rock. J. Geotech.
Engrg., ASCE, 118(6), 839–855.
Castelli, F., & Maugeri, M. (2002). Simplified nonlinear analysis for settlement prediction of pile
groups. J. Geotech. Geoenvir. Engrg., ASCE, 128(1), 76–84.
Castelli, F., Maugeri, M., & Motta, E. (1992). Analisi non lineare del cedimento di un Palo Singolo.
Rivista Italiana di Geotechnia, 26(2), 115–135.
CGS (1985). Canadian Foundation Engineering Manual. Part 2 (2nd ed.). Canadian Geotechnical
Society, Vancouver, Canada.
Chan, L.P. (1986). Application of Block Theory and Simulation Techniques to Optimum Design of
Rock Excavations. Ph.D. thesis, University of California, Berkeley, CA.
Chang, M.F., & Wong, I.H. (1987). Shaft friction of drilled piers in weathered rocks. Proc. 6th Int.
Cong. on Rock Mech., Montreal, 1, 313–318.
Chappel, B.A. (1974). Load distribution and deformational response in discontinua. Geotechnique,
24, 641–654.
Chen, E.P. (1989). A constitutive model for jointed rock mass with orthogonal sets of joints. J.
Applied Mech., ASME, 56, 25–32.
Cho, K.H. (2002). P-y Curves for Laterally Loaded Drilled Shafts Embedded in Soft Weathered
Rock. Ph.D. thesis, North Carolina State University, Raleigh, NC.
Cho, K.H., Clark, S.C, Keaney, B.D., Gabr, M.A., & Borden, R.H. (2001). Laterally loaded drilled
shafts embedded in soft rock. Transportation Research Record 1772, Paper No. 01–2998, 3–11.
Clarke, B.G. (1995). Pressuremeters in Geotechnical Design. Blackie Acdemic & Professional.
Coates, D.F. (1967). Rock Mechanics Principles. Energy Mines and Resources, Ottawa, Canada,
Monograph 874.
Coates, D.F., & Gyenge, M. (1966). Plate-load testing on rock for deformation and strength
properties. ASTM Special Technical Publication, 402, 19–35.
Coon, R.F., & Merritt, A.H. (1970). Predicting in situ modulus of deformation using rock quality
indices. Determination of the in Situ Modulus of Deformation of Rock, ASTM STP 477. 154–
173.
References 408
Cooper, S.S., & Ballard, R.F., Jr. (1988). Geophysical exploration for cavity detection in karst
terrain. Geotechnical Aspects of Karst Terrains: Exploration, Foundation Design and
Performance, and Remedial Measure, GSP No. 14, Ed.: N.Sitar, ASCE, New York, NY.
Couetdic, J.M., & Barron, K. (1975). Plate-load testing as a method of assessing the in situ strength
properties of Western Canadian coal. Int J. Rock Mech and Min. Sc., 12(10), 303–310.
Crofton, M.W. (1885). Probability. Encyclopedia Britannica. (9th ed.). 19, 768.
Cruden, D.M. (1977). Describing the size of discontinuities. Int. J. Rock Mech. Min. Sci. &
Cummings, D. (1990). Surface geophysical investigations for hazardous waste sites. Geophysical
Applications for Geotechnical Investigations, Eds: R.L.Paillet & W.R.Saunders, ASTM,
Philadelphia, PA, STP1101, 9–16.
Cundall, P.A. (1971). A computer model for simulating progressive large-scale movements in
blocky rock system. Proc. Symp. Int Soc. Rock Mech., Nancy, France, II–8.
Cundall, P.A. (1980). UDEC-A Generalized Distinct Element Programfor Modeling Jointed Rock.
Peter Cundall Associates, Report PCAR-1–80, U.S. Army, European Research Office, London.
Cundall, P.A. (1988). Formulation of a three-dimensional distinct element model—Part I. A
scheme to detect and represent contacts in a system composed of many polyhedral blocks. Int. J.
Rock Mech. Min. Sci. & Geomech Abstr., 25, 107–116.
Curran, J.H., & Ofoegbu, G.I. (1993). Modeling discontinuities in numerical analysis.
Comprehensive Rock Engrg.—Principle, Practice & Projects. Ed: J.A.Hudson, Pergamon Press,
1, 443–468.
Davisson, M.T. (1970). Lateral load capacity of piles. Highway Research Record, 333, 104–102.
Deere, D.U. (1964). Technical description of rock cores for engineering purposes. Rock Mech. and
Rock Engrg., 1, 107–116.
Deere, D.U. (1989). Rock Quality Designation (RQD) after Twenty Years. U.S. Army Corps of
Engineers Contract Report GL-89–1, Waterways Experiment Station, Viksburg, MS.
Deere, D.U., Hendron, A.J., Patton, F.D., & Cording, E.J. (1967). Design of surface and near
surface construction in rock. Failure and Breakage of Rock, Proc. 8th U.S. Symp. Rock Mech.,
Ed: C.Fairhurst, 237–302.
Deere, D.U., & Miller, R.P. (1966). Engineering Classiflcation and Index Properties for Intact
Rock. Tech. Rep. No. AFWL-TR-65-116, Air Force Weapons Lab, Kirtland Air Base, New
Mexico.
Deng, Z. (1996). The Application of Ground Penetrating Radar for Geological Characterization in
Three Dimensions at Gypsy Outcrop Site, Northeastern Oklahoma, USA. MS thesis, University
of Oklahoma, Norman, Oklahoma,
Dershowitz, W.S., Baecher, G.B., & Einstein, H.H. (1979). Prediction of rock mass deformability.
Proc. 4th Int. Cong. on Rock Mech., Montreal, Canada, 1, 605–611.
Dershowitz, W.S, & Herda, H.H. (1992). Interpretation of fracture spacing and intensity. Proc.
33rd U.S. Symp. on Rock Mech., Santa Fe, NM, 757–766.
Dershowitz, W.S, Lee., G., Geier, J., Hitchcock, S., & LaPointe, P. (1993). FracMan Version
2.306, Interactive discrete feature data analysis, geometric modeling, and exploration
simulation. User Documentation. Golder Associates Inc., Seattle, Washington.
Desai, C.S., & Christian, J.T. (1977). Numerical Methods in Geotechnical Engineering. McGraw-
Hill, NY.
Desai, C. S, Zaman, M.M., Lightner, J.G., & Siriwardane, H.J. (1984). Thin-layer element for
interface and joints. Int. J. Numerical & Analytical Methods in Geomech., 8, 19–43.
Digioia, A.M., Davidson, H.L., & Donovan, T.D. (1981). Laterally loaded drilled piers—A design
model. Drilled Piers and Caissons, Ed: M.W.O’Neill, 132–149.
Digioia, A.M., & Rojas-Gonzalez, L.F. (1993). Discussion on Analysis of laterally loaded shafts in
rock. J. Geotech. Engrg., ASCE, 119(12), 2014–2015.
Digioia, A.M., & Rojas-Gonzalez, L.F. (1994). Rock socket transmission line foundation
performance. IEE Transactions on Power Delivery, 9(3), 1570–1576.
References 409
Donald, I.B., Sloan, S.N., & Chiu, H.K. (1980). Theoretical analysis of rock-socketed piles. Proc.
Int. Conf. on Structural Found. on Rock, Sydney, 1, 303–316.
Donn, W.L., & Shimer, J.A. (1958). Graphic Methods in Structural Geology. Appleton Century
Crofts, New York.
Doruk, P. (1991). Analysis of the Laboratory Strength Data using the Original and Modified Hoek-
Brown Failure Criterion. MS thesis, Department of Civil Engineering, University of Toronto,
Toronto, Ontario, Canada.
Dykeman, P., & Valsangkar, A.J. (1996). Model studies of socketed caissons in soft rock. Can.
Geotech. J., Ottawa, Canada, 33(5), 747–759.
Ebisu, S., Aydan, O., Komura, S., & Kawamoto, T. (1992). Comparative study on various rock
mass characterization methods for surface structures. ISRM Symp.: Eurock ‘92, Rock
Characterization, Chester, UK, Ed: Hudson, J.A., 203–208.
Einstein, H.H., & Baecher, G.B. (1983). Probabilistic and statistical methods in engineering
geology (part I). Rock Mech. and Rock Engrg., 16(1), 39–72.
Einstein, H.H., Baecher, G.B., Veneziano, D., et al. (1979). Risk Analysis for Rock Slopes in Open
Pit Mines. Parts I–V, USBM Technical Report J0275015.
Einstein, H.H., & Hirschfeld, R.C. (1973). Model studies on mechanics of jointed rock. J. Soil
Mech Found. Div., ASCE, 99, 229–242.
Einstein, H.H., Veneziano, D., Baecher, G.B., & O’Reilly, K.J. (1983). The effect of discontinuity
persistence on rock slope stability. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 20(5), 227–
236.
Engel, R.L. (1988). Discussion of procedures for the determination of pile capacity. Transportation
Research Record 1169, TRB, Washington, DC, 54–61.
Ernst, H. (1995). East Milton Square-Osterberg Load Test. Massachusetts Highway Department.
Erwin, J.W., & Brown, R.A. (1988). Karstic foundation problems, Sunny Point Railroad.
Geotechnical Aspects of Karst Terrains: Exploration, Foundation Design and Performance, and
Remedial Measures, GSP No. 14, Ed.: N.Sitar, ASCE, New York, NY, 75–85.
Eurocode 7 (1994). Geotechnical Design.
Fellenius, B.H. (1980). The analysis of results from routine pile load tests. Ground Engrg., 6, 19–
31.
FHWA (1996a). Load and Resistance Factor Design (LRFD) for Highway Substructures. Manual
for NHI Course, Federal Highway Administration, Washington, DC.
FHWA (1996b). Design and Construction of Driven Pile Foundations. Publication No. FHWAHI-
97–013, Federal Highway Administration, Washington, DC.
Findlay, J.D., Brooks, N.J., Mure, J.N., & Heron W. (1997). Design of axially loaded piles—United
Kingdom practice. Design of Axially Loaded Piles—European Practice, Eds: De Cock &
Legrand, Balkema, Rotterdam.
Fishman, K.L., Derby, C.W., & Palmer, M.C. (1991). Verification for numerical modeling of
jointed rock mass using thin layer elements. Int. J. Numerical & Analytical Methods in
Geomech, 15, 61–70.
Fleming, W.G.K., Weltman, A.J., Randolph, M.F., & Elson, W.K. (1992). Piling Engineering.
Blackie.
Fossum, A.F. (1985). Effective elastic properties for a randomly jointed rock mass. Int. J. Roch
Mech Min. Sci. & Geomech. Abstr., 22(6), 467–470.
Frantzen, J., & Stratten, W.F. (1987). Final Report: p-y curve Data for Laterally Loaded Piles in
Shale and Sandstone. Kansas Department of Transportation, Report No. FNWA-KS-87–2.
Frechette, D.N., Walsh, K.D., & Houston, W.N., & Houston, S.L. (2001). Behavior of laterally
loaded drilled shaft groups at prototype scale. J. Geotech Geoenvir. Engrg., (in review, pre-
print).
Frechette, D.N., Walsh, K.D., & Houston, W.N. (2002). Review of design methods and parameters
for laterally loaded groups of drilled shafts. Deep Foundations 2002, Proc. Int. Deep Found.
Cong. 2002, GSP No. 116, ASCE, Eds: M.W.O’Neill & F.C.Townsend, 2, 1261–1274.
References 410
Gabr, M.A. (1993). Discussion on Analysis of laterally loaded shafts in rock. J. Geotech. Engrg.,
ASCE, 119(2) 2015–2018.
Gardner, W.S. (1987). Design of drilled piers in the Atlantic Piedmont. Foundations and
Excavations in Decomposed Rock of the Piedmont Province, Ed: R.E.Smith, GSP No. 9, ASCE,
62–86.
Gazioglu, S.M., & O’Neill, M.W. (1984). Evaluation of p-y relationships in cohesive soils.
Analysis and Design of Pile Found, Ed: J.R.Meyer. 192–213.
Gerrard, C.M. (1982a). Elastic models of rock masses having one, two and three sets of joints. Int.
J. Rock Mech Min. Sci. & Geomech. Abstr., 19, 15–23.
Gerrard, C.M. (1982b). Joint compliances as a basis for rock mass properties and the design of
supports. Int. J. Rock Mech. Min. Sci. & Geomech Abstr., 19, 285–305.
Gerrard, C.M. (1991). The equivalent elastic properties of stratified and jointed rock masses. Proc.
Int. Conf. on Computer Meth. and Advances in Geomech., Cairns, Eds: G.Beer. J.R.Brooker &
J.P.Carter, Balkema, Rotterdam, 333–337.
Gerrard, C.M., & Harrison, W.J. (1970). Circular loads applied to a cross-anisotropic half-space.
and Stresses and displacements in a loaded orthorhombic half-space. Technical Papers 8 and 9,
Division of Applied Geomechanics, Commonwealth scientific and Industrial Research
Organization, Australia (reproduced as Appendices A and B in Poulos, H.G. & Davis, E.H.,
Elastic Solutions for Soil and Rock Mechanics, John Wiley & Sons, NY)
Ghaboussi, J., Wilson, E.L., & Isenberg, J.R. (1973). Finite element for rock joints and interfaces.
J. Soil Mech. Found., ASCE, 99, 833–848.
Glos, G.H., III, & Briggs, O.H., Jr. (1983). Rock sockets in soft rock. J. Geotech. Engrg., ASCE,
109(4), 525–535.
Gnirk, P., Boyle, W.J., & Parrish, D.K. (1992). Quantifying geologic uncertainty in site
characterization.’ ISRM Symp.: Eurock ‘92, Rock Characterization, Chester, UK, Ed:
J.A.Hudson, 468–473.
Goeke, P.M., & Hustard, P.A. (1979). Instrumented drilled shafts in clay-shale. Symp. on Deep
Found., Ed.: E.M.Fuller, ASCE National Convention, Atlanta, Georgia, 149–164.
Goel, R.K., Jethwa, J.L., & Paithankar, A.G. (1995). Correlation between Barton’s Q and
Bieniaswki’s RMR-A new approach. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 33(2),
179–181.
Goodman, R.E. (1966). On the distribution of stresses around circular tunnels in non-homogeneous
rocks. Proc. 1st Cong on Rock Mech., Lisbon, Portugal, 2, 249–255.
Goodman, R.E. (1974). The mechanical properties of joints. Proc. 3rd Int. Cong on Rock Mech.,
Denver, CO, 1-A, 127–140.
Goodman, R.E. (1976). Methods in Geological Engineering. West Publishing Company.
Goodman, R.E. (1980). Introduction to Rock Mechanics. Wiley, New York.
Goodman, R.E. (1993). Engineering Geology-Rock in Engineering Construction. John Wiley &
Sons, Inc.
Goodman, R.E., & Bray, J.W. (1976). Toppling of rock slopes. Proc. Spec. Conf. on Rock Engrg.
for Found. and Slopes, Boulder, CO, 201–234.
Goodman, R.E., & Shi, G. (1985). Block Theory and its Application to Rock Engineering. Prentice-
Hall, Englewood Cliffs, NJ.
Goodman, R.E., Taylor, & Brekke, T.L. (1968). A model for the mechanics of jointed rock. J. Soil
Mech. Found., ASCE, 96, 637–659.
Goodman, R.E., Van, T.K., & Heuze, F.E. (1969). The measurement of rock deformability in
boreholes. Proc. 10th U.S. Symp. on Rock Mech., Austin, TX, 523–555.
Goodwin, J.W., Hayes, J.A., & Schmertmann, J.H. (1994). Report on Osterberg cell load testing,
SR 14 Bridge over the Black Warrier River, Greene County, Alabama, Loadtest, Inc.
Grant, R., Christian, J.T., & Vanmarcke, E.H. (1974). Differential settlement of buildings. J.
References 411
Griffiths, D.H. & King, R.F. (1981). Applied Geophysics for Geologists and Engineers. (2nd ed.).
Pergamon Press, Oxford.
Gunning, A.P. (1992). CCTV borehole surveying and its application to rock engineering. ISRM
Symp.: Eurock ‘92, Rock Characterization, Chester, UK, Ed: J.A.Hudson, 174–178.
Gupton, C., & Logan, T. (1984). Design guidelines for drilled shafts in weak rocks of south
Florida. Proc. South Florida Annual ASCE Meeting, ASCE.
Habibagahi, K., & Langer, J.A. (1983). Horizontal subgrade modulus of granular soils. Laterally
Loaded Deep Foundations—Analysis and Performance. Eds: J.A.Langer, E.T.Mosley, & C.D.
Thompson, ASTM STP 835.
Hansen, B.J. (1961). The ultimate resistance of rigid piles against transversal forces. Geoteknist
Institut. Bull. No. 12, Copenhagen.
Harrison, J.P. (1992). Fuzzy objective functions applied to the analysis of discontinuity orientation
data. ISRM Symp.: Eurock ‘92, Rock Characterization, Chester, UK, Ed: J.A.Hudson, 25–30.
Harrison, J.P. (1999). Selection of the threshold value in RQD assessments. Int. J. Rock Mech Min.
Sci., 36, 673–685.
Hart, R., Cundall, P.A., & Lemos, J. (1988). Formulation of a three-dimensional distinct element
model—Part II. Mechanical calculations for motion and interaction of a system composed of
many polyhedral blocks. Int. J. Rock Mech Min. Sci & Geomech. Abstr., 25, 117–126.
Hassan, K.M., & O’Neill, M.W. (1997). Side load transfer mechanism in drilled shafts in soft
argillaceous rock. J. Geotech, Engrg., ASCE, 123(2), 272–280.
Hassan, K.M., O’Neill, M.W., Sheikh, S.A., & Ealy, C.D. (1997). Design method for drilled shafts
in soft argillaceous rock. J. Geotech. Engrg., ASCE, 123(3), 272–280.
Heath, W.E. (1995). Drilled pile foundations in porous, pinnacled carbonate rock. Proc. 5th
multidisciplinary Conf. on Sinkholes and the Engrg. and Environ. Impacts of Karst, Ed: B.F.
Beck, 371–374.
Heuze, F.E. (1980). Scale effects in the determination of rock mass strength and deformability.
Rock Mech., 12,167–192.
Heuze, F.E. (1984). Suggested method for estimating the in-situ modulus of deformation of rock
using the NX-borehole jack. Geotech. Testing J., 7(4), 205–210.
Heuze, F.E., & Amadei, B. (1985). The NX borehole jack—A lesson in trials and errors. Int. J.
Rock Mech. Min. Sci. & Geomech Abstr., 22(2), 105–112.
Heuze, F.E., & Barbour, T.G. (1982). New models for rock joints and interfaces. J. Geotech.
Engrg., ASCE, 108, 757–776.
Hills, E.S. (1972). Elements of Structural Geology. (2nd ed.). London, Chapman & Hall.
Hobbs, B.E. (1976). An Outline of Structural Geology. New York, Wiley.
Hoek, E. (1983). Strength of jointed rock masses—23rd Rankine Lecture. Geotechnique, 33(3),
187–223.
Hoek, E. (1990). Estimating Mohr-Coulomb friction and cohesion values from the Hoek-Brown
failure criterion. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 27(3), 227–229.
Hoek, E. (1994). Strength of rock and rock masses. News J. of IRSM, 2(2), 4–16.
Hoek, E., & Bray, J.W. (1981). Rock Slope Engineering. (3rd ed.). Institution of Mining and
Metallurgy, London.
Hoek, E., & Brown, E.T. (1980). Empirical strength criterion for rock masses. J. Geotech. Engrg.,
ASCE, 106, 1013–1035.
Hoek, E., & Brown, E.T. (1988). The Hoek-Brown criterion—a 1988 update. Proc. 15th Can. Rock
Mech. Symp., University of Toronto, Canada, 31–38.
Hoek, E., & Brown, E.T. (1997). Practical estimates of rock mass strength. Int J. Rock Mech. Min.
Sci., 34(8), 1165–1186.
Hoek, E., Kaiser, P.K., & Bawden, W.F. (1995). Support of Underground Excavations in Hard
Rock. A.A.Balkema, Rotterdam.
Hoek, E., Wood, D., & Shah, S. (1992). A modified Hoek-Brown criterion for jointed rock masses.
Proc. Rock Characterization Symp. ISRM: Eurock’92, Ed: J.A.Hudson, 209–214.
References 412
Hoit, M., Hays, C., & McVay, M.C. (1997). The Florida Pier Analysis Program—methods and
models for pier analysis and design. Design and analysis of foundations and sand liquefaction
TRB Rec. No. 1569. . Transp. Res. Board, Washington, DC, 1–8.
Holden, J.C. (1984). Construction of Bored Piles in Weathered Rocks. Technical Report 69, Road
Construction Authority of Victoria, Melbourne, Australia.
Horvath, R.G. (1982). Drilled Piers Socketed into Weak Shale—Methods of Improving
Performance. Ph.D. Dissertation, University of Toronto, Toronto, Ontario, Canada.
Horvath, R.G., & Kenney, T.C. (1979). Shaft resistance of rock-socketed drilled piers. Proc.
American Society of Civil Engineers Annual Convention, Atlanta, Preprint 3698.
Horvath, R.G., Kenney, T.C., & Kozicki, P. (1983). Method of improving the performance of
drilled piers in weak rock. Can. Geotech. J., Ottawa, Canada, 20(4), 758–772.
Horvath, R.G., Kenney, T.C., & Trow, W.P. (1980). Results of tests to determine shaft resistance of
rock-socketed drilled piers. Proc. Int. Conf. on Structural Found. on Rock, Sydney, 1, 349–361.
Hummert, J.B., & Cooling, T.L. (1988). Drilled pier test, Fort Collins Colorado. Proc. 2nd Int.
Conf. on Case Histories in Geotech. Engrg., St. Louis, Missouri, 3, 1375–1382.
Hu, Yushan (1985). The analysis of laterally-loaded rock-socketed bored test piles. Selected Papers
from the Chinese J. of Geotech. Engrg.—1985. Ed: Yang H.Huang. 113–122.
ISRM (1978). Suggested methods for the quantitative description of discontinuities in rock masses.
International Society for Rock Mechanics, Commission on Standardization of Laboratory and
Field Tests. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 15, 319–368.
ISRM (1979a). Suggested methods for determining the uniaxial compressive strength and
deformability of rock materials. International Society for Rock Mechanics, Commission on
Standardization of Laboratory and Field Tests. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr.,
16, 135–140.
ISRM (1979b). Suggested methods for determining in situ deformability of rock. International
Society for Rock Mechanics, Commission on Standardization of Laboratory and Field Tests.
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 16, 195–214.
ISRM (1987). Suggested methods for deformability determination using a flexible dilatometer.
International Society for Rock Mechanics, Commission on Standardization of Laboratory and
Field Tests. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 24, 123–134.
Ivanova, V. (1998). Geological and Stochastic Modeling of Fracture Systems in Rocks. PhD thesis,
Massachusetts Institute of Technology, Cambridge, MA.
Jaeger, J.C. (1960). Shear failure of anisotropic rocks. Geological Magazine, 97, 65–72.
Jaeger, J.C. (1971). Friction of rocks and the stability of rock slopes, Rankine Lecture.
Geotechnique, London, UK, 21, 97–134.
Jaeger, J.C., & Cook, N.G.W. (1979). Fundamentals of Rock Mechanics. (3rd ed.). Chapman &
Hall, London.
Jamiolkowski, M., & Garrassino, A. (1977). Soil modulus for laterally loaded piles. Spec. Sess. No.
10, 9th Int. Conf. on Soil Mech. Found. Engrg., Tokyo, 87–92.
John, K.M. (1970). Civil engineering approach to evaluate strength and deformability of closely
jointed rock. Rock Mech.—Theory and Practice, Proc. 14th U.S. Symp. on Rock Mech., 69–80.
John, M. (1972). The Influence of Length to Diameter Ratio on Rock Properties in Uniaxial
Compression: A Contribution to Standardization in Rock Mechanics Testing. Rep. S. Afr. CSIR,
No. ME 1083/5.
Johnston, I.W. (1985). Strength of intact geomechanical materials. J. Geotech. Engrg., ASCE, 111,
730–749.
Johnston, I.W. (1992). New Developments in the Predictions of Side Resistance of Piles in Soft
Rock. Civil Engineering Department, Monash University, Clayton, Victoria, Australia.
Johnston, I.W., & Choi, I.K. (1985). Failure mechanism of foundations in soft rock. Proc. 1st Int.
Conf. on Soil Mech. and Found. Engrg., San Francisco, 3, 1397–1400.
References 413
Johnston, I.W., Donald, I.B., Bennett, A.G., & Edwards, J. (1980). The testing of large diameter
pile rock sockets with a retrievable test rig. Proc. 3rd Australia-New Zealand Conf on
Geomech., Wellington, 1, 105–108.
Jubenville, M.D., & Hepworth, R.C. (1981). Drilled pier foundations in shale—Denver Colorado
area. Drilled Piers and Caissons, Proc. Session at the ASCE National Convention, St. Louis,
Missouri, 66–81.
Kachanov, M. (1980). Continuum model of medium with cracks. J. Engrg. Mech. Div., ASCE,
106(5), 1039–1051.
Kalamaras, G.S., & Bieniawski, Z.T. (1993). A rock mass strength concept for coal seams. Proc.
12th Conf. Ground Control in Mining, Morgantown, 274–283.
Karzulovic, A., & Goodman, R.E. (1985). Determination of principal joint frequencies. Int. J. Rock
Mech. Min. Sci. & Geomech. Abstr., 22(6), 471–473.
Katzenbach, R., Arslan, U., & Holzhauser, J. (1998). Group efficiency of a large pile group in rock.
Deep Foundations on Bored and Auger Piles—BAP III, Proc. 3rd Int. Geotech. Seminar on
Deep Found. on Bored and Auger Piles,, Ghent Belgium, 223–229.
Kawamoto, T., Ichikawa, Y., & Kyoya, T. (1988). Deformation and fracturing behavior of
discontinuous rock mass and damage mechanics theory. Int. J. Numerical & Analytical Methods
in Geomech., 12, 1–30.
Kearey, P., & Brooks, M. (1991). An Introduction to Geophysical Exploration. (2nd ed.). Blackwell
Scientific Publications, London, UK.
Kodikara, J.K., Johnston, I.W., & Haberfield, C.M. (1992). Analytical predictions of shear
resistance of piles in rock. Proc. 6th Australia-New Zealand Conf. on Geomech., Christchurch,
New Zealand, 156–162.
Kraft, L.M., Ray, R.P., & Kagawa, T. (1981). Theoretical t-z curves. J. Geotech. Engrg., ASCE,
107(11), 1543–1561.
Kubo, J. (1965). Experimental study of the behavior of laterally loaded piles. Proc. 6th Int. Conf.
on Soil Mech. and Found. Engrg., 2, 275–279.
Kulatilake, P.H.S.W. (1985a). Fitting Fisher distributions to discontinuity orientation data. J. of
Geological Education, 33(5), 266–269.
Kulatilake, P.H.S.W. (1985b). Estimating elastic constants and strength of discontinuous rock. J.
Geotech. Engrg., ASCE, 111, 847–864.
Kulatilake, P.H.S.W. (1986). Bivariate normal distribution fitting on discontinuity orientation
clusters. Mathematical Geology, 18(2), 181–195.
Kulatilake, P.H.S.W. (1988). State-of-the-art in stochastic joint geometry modeling. Key Questions
in Rock Mech.: Proc. 29th U.S. Symp. on Rock Mech., 215–299.
Kulatilake, P.H.S.W., Wu, T.H., & Wathugala, D.N. (1990). Probabilistic modeling of joint
orientation. Int. J. Numerical & Analytical Methods in Geomech., 14, 325–350.
Kulatilake, P.H.S.W. (1993). Application of probability and statistics in joint network modeling in
three dimensions. Proc. Conf. on Probabilistic Methods in Geotech. Engrg., Lanberra,
Australia, 63–87.
Kulatilake, P.H.S.W., Ucpirti, H., Wang, S., Radberg, G, & Stephansson, O. (1992). Use of the
distinct element method to perform stress analysis in rock with non-persistent joints and to study
the effect of joint geometry parameters on the strength and deformability of rock. Rock Mech.
and Rock Engrg., 25(4), 253–274.
Kulatilake, P.H. S.W., Wang, S., & Stephansson, O. (1993). Effect of finite size joints on the
deformability of jointed rock in three dimensions. Int. J. Rock Mech. Min. Sci. & Geomech.
Abstr., 30(5), 479–501.
Kulatilake, P.H. S.W., Wathugala, D.N., & Stephansson, O. (1993). Stochastic three dimensional
joint size, intensity and system modeling and a validation to an area in Stripa Mine, Sweden.
Soils and Found., 33(1), 55–70.
Kulatilake, P.H.S.W., & Wu, T.H. (1984a). Sampling bias on orientation of discontinuities. Rock
Mech. and Rock Engrg., 17, 243–253.
References 414
Kulatilake, P.H. S.W., & Wu, T.H. (1984b). The density of discontinuity traces in sampling
windows. Int. J. Rock Mech. Min. Sci. & Geomech, Abstr., 21(6), 345–347.
Kulatilake, P.H. S.W., & Wu, T.H. (1984c). Estimation of mean trace length of discontinuities.
Rock Mech. and Rock Engrg., 17, 215–232.
Kulhawy, F.H. (1978). Geomechanical model for rock foundation settlement. J. Geotech. Engrg.,
ASCE, 104(2), 211–227.
Kulhawy, F.H., & Carter, J.P. (1992). Settlement and bearing capacity of foundations on rock
masses. Engineering in Rock Masses, Ed: F.G. Bell, Butterworth-Heinemann, Oxford, UK,
231–246.
Kulhawy, F.H., & Goodman, R.E. (1980). Design of foundations on discontinuous rock. Int. Conf.
on Struc. Found. on Rock, Ed: P.J. N. Pells, Balkema, Rotterdam, 209–220.
Kulhawy, F.H., & Goodman, R.E. (1987). Foundation in rock. Ground Engineering Reference
Book, Ed. F.G. Bell, Butterworth, London, 55/1–55/13.
Kulhawy, F.H. (1991). Drilled shaft foundations. Foundation Engineering Book. Ed: H.Y. Fang,
Van Nostrand Reinhold, NY, 537–552.
Kulhawy, F.H., & Phoon, K.K. (1993). Drilled shaft side resistance in clay soil to rock. Proc. Conf.
on Design and Performance of Deep Found: Piles and Piers in Soil and Soft Rock. Geotechnical
Special Publication No. 38, ASCE, 172–183.
Ladanyi, B., & Archambault, G. (1970). Simulation of shear behavior of a jointed rock mass. Proc.
11th U.S. Symp. on Rock Mech., AIME, New York, 105–125.
Lama, R.D., & Vutukuri, V.S. (1978). Handbook on mechanical properties of rocks. Trans. Tech.
Publ., 4, 317–399.
Landanyi, B., & Roy, A. (1971). Some aspects of the bearing capacity of rock mass. Proc. 7th
Canadian Symp. Rock Mech., Edmonton.
LaPointe, P.R., Wallmann, P.C., & Dershowitz. W.S. (1993). Stochastic estimation of fracture size
through simulated sampling. Int J. Rock Mech. Min. Sci. & Geomech. Abstr., 30(7), 1611–1617.
Laslett, G.M. (1982). Censoring and edge effects in areal and line transect sampling of rock joint
traces. Mathematical Geology, 14(2), 125–140.
Lauffer, H. (1958). Gebirgsklassifizierung für den Stollenbau. Geol Bauwesen, 24(1), 46–51.
Lee, R.D. (1961). Testing mine floors. Colliery Engrg., 38(4), 255–261.
Lee, Y.H., Carr, J.R., Barr, D.J., & Haas, C.J. (1990). The fractal dimension as measure of the
roughness of rock discontinuity profiles. Int, J. Rock Mech. Min. Sci. & Geomech. Abstr., 27(6),
453–464.
Lee, J-S, Einstein, H.H., & Veneziano, D. (1990). Stochastic and centrifuge modeling of jointed
rock: Part III—Stochastic and topological fracture geometry model. Grant No. AFOSR-87–
0260, MIT.
Lemos, J.V., Hart, R.D., & Cundall, P.A. (1985). A generalized distinct element program for
modeling jointed rock mass. Proc. Symp. on Fundamentals of Rock Joints, Bjorkliden, Sweden,
335–343.
Leong, E.C., & Randolph, M.F. (1994). Finite element modeling of rock-socketed piles. Int. J.
Numerical & Analytical Methods in Geomech., 18, 25–47.
Leung, C.F. (1996). Case studies of rock-socketed piles. Geotech. Engrg. J., Southeast Asia
Geotechnical Society, 27(1), 51–67.
Leung, C.F., & Ko, H-Y (1993). Centrifuge model study of piles socketed in soft rock. Soils and
Found, Tokyo, Japan, 33(3), 80–91.
LOADTEST (2001). Contents in the web site of LOADTEST, Inc.—http://www.loadtest.com/.
Lorig, L.J., Brady, B.H. G., & Cundall, P.A. (1986). Hybrid distinct element-boundary element
analysis of jointed rock. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 23, 303–312.
Lunard, P., Froldi, P., & Fornari, E. (1994). Rock mechanics investigations for rock slope stability
assessment. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 31, 323–345.
Madhav, M.R., & Rao, N.S. V.K. (1971). Model for machine-pile foundation-soil system. J. Soil
Mech. Found. Div., ASCE, 97(1), 295–299.
References 415
Mahtab, M.A., Bolstad, D.D., Alldredge, J.R., & Shanley, R.J. (1972). Analysis of Fracture
Orientations for Input to Structural Models of Discontinuous Rock. US Bur. Mines Rep. Invest.
7669.
Mahtab, M.A., & Yegulalp, T.M., (1984). A similarity test for grouping orientation data in rock
mechanics. Proc. 25th U.S. Symp. on Rock Mech., 495–502.
Marinos, P., & Hoek, E. (2001). Estimating the geotechnical properties of heterogeneous rock
masses such as flysch. Bull Engrg. Geol Env., 60, 85–92.
Matlock, H. (1970). Correlations for design of laterally loaded piles in soft clay. Proc. 2nd Offshore
Technology Conf., Offshore Technology Conference, Houston, Tex., 1, 577–594.
Matlock, H., & Foo, S.H. C. (1976). Full-scale lateral load tests of pile groups. Discussion, J.
Matlock, H., & Reese, L.C. (1960). Generalized solutions for laterally loaded piles. J. Soil Mech.
Fotmd Div., ASCE, 86(5), 63–91.
Matlock, H., & Ripperberger, E.A. (1958). Measurement of soil pressure on a laterally loaded pile.
Proc. ASTM, 58, 1245–1259.
Mattes, N.S., & Poulos, H.G. (1969). Settlement of single compressive piles. J. Soil Mech. Found.
Div., ASCE, 95(1), 189–207.
Mauldon, M. (1994). Intersection probabilities of impersistent joints. Int. J. Rock Mech. Min. Sci. &
Geomech. Abstr., 31(2), 107–115.
Mauldon, M. (1998). Fracture sampling on a cylinder: From scanline to boreholes and tunnels.
Rock Mech. and Rock Engrg., 30(3), 129–144.
Mauldon, M., & Mauldon, J.G. (1997). Estimating mean fracture trace length and density from
observations in convex windows. Rock Mech. and Rock Engrg., 31(4), 201–216.
McMillan, P., Blair, A.R., & Nettleton, I.M. (1996). The use of down-hole CCTV for collection of
quantitative discontinuity data. Proc. Int. Conf. on Advances in Site Investigation Practice,
London, 312–323.
McVay, M. C, Twonsend, F. C, & Williams, R.C. (1992). Design of socketed drilled shafts in
limestone. J. Geotech. Engrg., ASCE, 118(10), 1626–1637.
McWilliams, P. C, Kerkering, J.C., & Miller, S.M. (1990). Fractal characterization of rock fracture
roughness for estimating shear strength.’ Proc. Int. Conf. on Mech. of Jointed and Faulted Rock,
Vienna, Aurtria, Ed: H.P. Rossmanith, Balkema, Rotterdam, 331–336.
Medhurst, T.P., & Brown, E.T. (1996). Large scale laboratory testing of coal. Proc. 7th Australian-
New Zealand Conf on Geomech., Canberra, Australia, 203–208.
Meigh, A.C., & Wolski, W. (1979). Design parameters for weak rocks. Proc. 7th European Conf.
on Soil Mech. and Found. Engrg., Brighton, British Geotechnical Society, 5, 57–77.
Meyer, T. (1999). Geologic Stochastic Modeling of Rock Fracture Systems Related to Crustal
Faults. MS thesis, Massachusetts Institute of Technology, Cambridge, MA.
Middendorp, P., Bermingham, P., & Kuiper, B. (1992). Statnamic load testing of foundation piles.
Proc. 4th Int. Conf on Application of Stress wave Theory to Piles, Balkema, Rotterdam, 581–
588.
Miller, S.M. (1983). A statistical method to evaluate homogeneity of structural populations.
Mathematical Geology, 15(2), 317–328.
Moh, Z.C., Yu, K., Toh, P.H., & Chang, M.F. (1993). Base and shaft resistance of bored piles
founded in sedimentary rocks. Proc. 11th Southeast Asian Geotech. Conf., Singapore, 571–576.
Mostyn, G.R., & Li, K.S. (1993). Probabilistic slope analysis—State-of-play. Proc. Conf. on
Probabilistic Methods in Geotech.. Engrg., Canberra, Australia, 89–109.
Moulton, L.K., GangaRao, H.V. S., & Halvorsen, G.T. (1985). Tolerable Movement Criteria for
Highway Bridges. Report No. FHWA/RD-85/107, Federal Highway Administration, U.S.
Department of Transportation, U.S. Government printing Office, Washington, DC.
Murchison, J.M., & O’Neill, M.W. (1984). Evaluation of p-y relationships in cohesionless soils.
Analysis and Design of Pile Found., Ed: J.R. Meyer. 174–191.
References 416
Murphy, D.K., et al. (1976). The LG-2 underground power house. Proc. RETC Conf., Las Vegas,
NV, 515–533.
Navy (1982). Foundations and Earth Structures. NAWAC DM-7.2, Naval Facilities Engineering
Command, US Government Printing Office, Washington, DC.
Ng, P.C. F., Pyrah, I. C, & Anderson, W.F. (1997). Assessment of three interface elements and
modification of the interface element in CRISP90. Computers and Geotechnics, 21(4), 315–339.
Oda, M. (1982). Fabric tensor for discontinuous geological materials. Soils and Found, 22, 96–108.
Oda, M., Suzuki, K., & Maeshibu, T. (1984). Elastic compliance for rock-like materials with
random cracks. Soils and Found., 24, 27–40.
Oda, M. (1988). An experimental study of the elasticity of mylonite rock with random cracks. Int J.
Rock Mech. Min. Sci. & Geomech. Abstr., 25, 59–69.
Odling, N.E. (1994). Natural fracture profiles, fractal dimension and joint roughness coefficients.
Rock Mech. and Rock Engrg., 27(3), 135–153.
Ogata, N., & Gose, S. (1995). Sloping rock layer foundation of bridge structure. Rock Foundation,
Proc. Int. Workshop on Rock Foundation, Tokyo, Japan, Eds: Yoshinaka & Kikuchi, Balkema,
Rotterdam.
Oliveira, R. & Charrua Graca, J. (1987). In situ testing of rocks. Ground Engineer ‘s Reference
Book, Ed: F.G. Bell.
O’Neill, M.W. (1998). Applications of large-diameter bored piles in the United States. Deep
Foundations on Bored and Auger Piles—BAP III, Proc. 3rd Int. Geotech. Seminar on Deep
Found. on Bored and Auger Piles, Ghent, Belgium, 3–19.
O’Neill, M.W., Brown, D.A., Townsend, F.C., & Abar, N. (1997). Innovative load testing of deep
foundations. Transportation Research Record 1569, Transportation Research Board,
Washington, DC, 17–25.
O’Neill, M.W., & Hassan, K.M. (1994). Drilled shafts: effects of construction on performance and
design criteria. Proc. Int. Conf. on Design and Construction of Deep Found., FHWA,
Washington, DC, 1, 137–187.
O’Neill, M.W., & Reese, L.C. (1999). Drilledshafts: Construction Procedures and Design
Methods. Report prepared for U.S. Department of Transportation and Federal Highway
Administration, Report No. FHWA-IF-99–025.
O’Neill, M.W., Townsend, F.C., Hassan, K.M., Buller, A., & Chan, P.S. (1996). Load Transfer for
Drilled Shafts in Intermediate Geomaterials. FHWA-RD-95–171, FHWA, U.S. Department of
Transportation
Orpwood, T.G., Shaheen, A.A., & Kenneth, R.P. (1989). Pressuremeter evaluation of glacial till
bearing capacity in Toronto, Canada. Foundation Engineering: Current Principle and
Practices. Ed: F.H. Kulhawy, ASCE, 1,16–28.
Osterberg, J.O. (1984). A new simplified method for load testing drilled shafts. Found. Drilling,
ADSC, 23(6).
Osterberg, J.O. (1989). New load cell testing device. Proc. 14th Annual Conf, 17–28.
Osterberg, J.O. (1998). The Osterberg load test method for drilled shafts and driven piles. Proc. 7th
Int. Conf on Piling and Deep Found, Vienna Austria.
Osterberg, J.O., & Gill, S.A. (1973). Load transfer mechanisms for piers socketed in hard soils or
rock. Proc. 9th Canadian Sym. on Rock Mech., Montreal, 235–262.
Pabon, G., & Nelson, P.P. (1993). Behavior of instrumented model piers in manufactured rock with
a soft layer. Geotechnical Special Publication No. 39, ASCE, 260–276.
Pahl, P.J. (1981). Estimating the mean length of discontinuity traces. Int. J. Rock Mech. Min. Sci. &
Geopmech. Abstr., 18, 221–228.
Palmer, L.A., & Thompson, J.B. (1948). The earth pressure and deflection along the embedded
lengths of piles subjected to lateral thrusts. Proc. 2nd Int. Conf. on Soil Mech and Found.
Engrg., Rotterdam., 5, 156–161.
Palmstrom, A. (1982). The volumetric joint count—a useful and simple measure of the degree of
rock jointing. Proc. 41st Int. Congress Int. Ass. Eng. Geol., Delphi, 5, 221–228.
References 417
Palmstrom, A. (1985). Application of the volumetric joint count as a measure of rock mass jointing.
Proc. Int. Symp. on Fundamentals of Rock Joints, Bjorkliden, Sweden, 103–110.
Palmstrom, A. (1986). A general practical method for identification of rock masses to be applied in
evaluation of rock mass stability conditions and TBM boring progress. Proc. Conf. on
Fjellsprengningsteknikk, Bergmekanikk, Geoteknikk, Oslo, Norway, 31,1–31.
Palmstrom, A. (1996). RMi—A system for characterizing rock mass strength for use in rock
engineering. J. of Rock Mech. and Tunneling Tech., 1(2), 69–108.
Pan, X.D., & Hudson, J.A. (1988). A simplified three-dimensional Hoek-Brown yield criterion.
Rock Mechanics and Power Plants, Ed: M. Romana, Balkema, Rotterdam, 95–103.
Pande, G.N., Beer, G., & Williams, J.R. (1990). Numerical Methods in Rock Mechanics. John
Wiley and Sons Ltd, England.
Pande, G.N., & Xiong, W. (1982). An improved multilaminate model of jointed rock masses.
Numerical Models in Geomech., Eds: R. Dungar, G.N. Pande & J.A. Studer, Balkema,
Rotterdam, 218–226.
Patton, F.D. (1966). Multiple modes of shear failure in rock. Proc. 1st Int. Cong. on Rock Mech.,
Lisbon, 1, 509–515.
Peck, R.B., Hanson, W.E., & Thornburn, T.H. (1974). Foundation Engineering. (2nd ed.). John
Wiley and Sons, New York.
Pells, P.J. N., Rowe, R.K., & Turner, R.M. (1980). An experimental investigation into side shear
for socketedpiles in sandstone. Proc. Int. Conf on Structural Found. on Rock, Sydney, 1,291–
302.
Pells, P.J. N., & Turner, R.M. (1979). Elastic solutions for the design and analysis of rock socketed
piles. Can. Geotech. J., Ottawa, Canada, 16, 481–487.
Pells, P.J. N., & Turner, R.M. (1980). End-bearing on rock with particular reference to sandstone.
Proc. Int. Conf on Structural Found. on Rock, Sydney, 1,181–190.
Petit, J.-P., Massonnat, G., Pueo, F., & Rawnsley, K. (1994). Rapport de forme des fractures de
mode 1 dans les roches stratifiees: Une etude de cas dans le Bassin Permian de Lodeve (France).
Bulletin du Centre de Recherches Elf Exploration Production, 18, 211–229.
Phillips, F.C. (1971). The Use of Stereographic Projection in Structural Geology. (3rd ed.). Edward
Arnold, London.
Piteau, D.R. (1970). Geological factors significant to the stability of slopes cut in rock. Symp. on
Planning Open Pit Mines, South African Inst. of Mining and Metallurgy, Johannesburg, 33–53.
Piteau, D.R. (1973). Characterizing and extrapolating rock joint properties in engineering practice.
Rock Mech. Supplement, 2, 5–31.
Poon, C.Y., Sayles, R.S., & Jones, T.A. (1992). Surface measurement and fractal characterization
of naturally fractured rocks. J. Phys. D: Applied Phys., 25, 1269–1275.
Poulos, H.G. (1971a). Behavior of laterally loaded piles: I-Single piles. J. Soil Mech. Found. Div.,
ASCE, 97(5), 711–731.
Poulos, H.G. (1971b). Behavior of laterally loaded piles: II-Pile groups. J. Soil Mech. Found Div.,
ASCE, 97(5), 733–751.
Poulos, H.G. (1972). Behavior of laterally loaded piles; IH-Socketed piles. J. Soil Mech. Found.
Div., ASCE, 97(4), 341–360.
Poulos, H.G. (1973). Load-deflection prediction for laterally loaded piles. Aust. Geotech. J.,
Australia, G3, 1–8.
Poulos, H.G. (1979). Group factors for pile deflection estimation. J. Geotech. Engrg., ASCE,
105(12), 1489–1509.
Poulos, H.G. (1989). Pile behavior-theory and application. Geotechnique, 39(3), 365–415.
Poulos, H.G. (2001). Pile foundations. Geotechnical and Geoenvironmental Engineering
Handbook, Ed: R.K. Rowe, Kluwer Academic Publishers, 261–304.
Poulos, H.G. & Davis, E.H. (1974). Elastic Solutions for Soil and Rock Mechanics. John Wiley and
Sons, NY.
References 418
Poulos, H.G. & Davis, E.H. (1980). Pile Foundation Analysis and Design. John Wiley and Sons,
NY.
Poulos, H.G., & Madhav, M.R. (1971). Analysis of the movement of battered piles. Proc. 1st
Australia-New Zealand Conf. on Geomech, 268–275.
Poulos, H.G., & Mattes, N.S. (1969). The behavior of axially-loaded end-bearing piles.
Pratt, H.R., Black, A.D., Brown, W.S., & Brace, W.F. (1972). The effect of specimen size on the
mechanical properties of unjointed diorite. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 9,
513–529.
Price, N.J. (1966). Fault and Joint Development in Brittle and Semi-Brittle Rock. Oxford,
Pergamon.
Priest, S.D. (1993). Discontinuity Analysis for Rock Engineering. Chapman & Hall.
Priest, S.D., & Hudson, J. (1976). Discontinuity spacing in rock. Int. J. Rock Mech. Min. Sci. &
Priest, S.D., & Hudson, J. (1981). Estimation of discontinuity spacing and trace length using
scanline surveys. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 18, 13–197.
Radhakrishnan, R., & Leung, C.F. (1989). Load transfer behavior of rock-socketed piles. J.
Ramamurthy, T. (1986). Stability of rock mass, Eighth Indian Geotech. Soc. Annual Lecture.
Indian Geotech. J., 16, 1–73.
Ramamurthy, T. (1993). Strength and modulus responses of anisotropic rocks. Comprehensive
Rock Engineering-Principle, Practice & Projects. Ed: J.A. Hudson, Pergamon Press, 1,313–
329.
Ramamurthy, T., & Arora, V.K. (1991). A simple stress-strain model for jointed rocks. Proc. 7th
Int. Cong. on Rock Mech., Anchen, Ed: W. Wittke, Balkema, Rotterdam, 1,323–326.
Ramamurthy, T., Rao, G.V., & Rao, K.S. (1985). A strength criterion for rocks. Proc. Indian
Geotech. Conf, Roorkee, 1, 59–64.
Randolph, M.F. (1981). The response of flexible piles to lateral loading. Geotechnique, 31(2), 247–
259.
Randolph, M.F. (1994). Design methods for pile groups and piled rafts. Proc. 13th Int. Conf on Soil
Mech. and Found. Engrg., New Delhi, 4, 61–82.
Randolph, M.F., & Houlsby, G.T. (1984). The limiting pressure on a circular pile loaded laterally
in cohesive soil. Geotechnique, 34(4), 613–623.
Randolph, M.F. & Wroth, C.P. (1978). Analysis of deformation of vertically loaded piles. J.
Reese, L.C. (1997). Analysis of laterally loaded piles in weak rock. J. Geotech. Geoenvir. Engrg.,
ASCE, 101(7), 633–649.
Reese, L. C., & Cox, W.R. (1969). Soil behavior from analysis of tests on uninstrumented piles
under lateral loading. ASTM, STP 444, 160–176.
Reese, L. C., Cox, W.R., & Koop, F.D. (1974). Analysis of laterally loaded piles in sand. Proc. 6th
Offshore Technology Conf., Houston, TX, 2,473–483.
Reese, L.C., & Matlock, H. (1956). Non-dimensional solutions for laterally loaded piles with soil
modulus proportional to depth. Proc. 8th Texas Conf. on Soil Mech. and Found. Engrg., Austin,
Tex.
Reese, L.C., & O’Neill, M.W. (1987). Drilled Shafts: Construction Procedures and Design
Methods. Design Manual, U.S. Department of Transportation, Federal Highway Administration,
Mclean, VA.
Reese, L.C., & Van Impe, W.F. (2001). Single Piles and Pile Groups under Lateral Loading. A.A.
Balkema, Rotterdam.
Reese, L.C., & Welch, R.C. (1975). Lateral loadings of deep foundations in stiff clay. J. Geotech.
Engrg., ASCE, 101(7), 633–649.
References 419
Reynolds, R.T., & Kaderabek, T.J. (1980). Miami Limestone Foundation Design and Construction.
ASCE, New York, N.Y.
Rhodes, G.W., Stephenson, R.W., & Rockaway, J.D. (1973). Plate bearing tests on coal underclay.
Proc. 19th U.S. Symp. on Rock Mech., Stateline, NV, 2, 16–27.
Roberds, W. I, & Einstein, H.H. (1978). Comprehensive model for rock discontinuities. J. Geotech.
Engrg., ASCE, 104(5), 553–569.
Roberds, W.J., Iwano, M., & Einstein, H.H. (1990). Probabilistic mapping of rock joint surfaces.
Proc. Int. Symp. on Rock Joints, Loen, Norway, Eds: N. Barton & O. Stephanson, Balkema,
Rotterdam, 681–691.
Robertson, A. (1970). The interpretation of geologic factors for use in slope theory. Proc. Symp. on
the Theoretical Background to the Planning of Open Pit Mines, Johannesburg, South Africa,
55–71.
Robertson, P.K., Huges, J.M. O., Camanella, R.G., & Sy, A. (1982). Design of laterally loaded
displacement piles using a driven pressuremeter. Laterally Loaded Deep Found, ASTM STP
835, Kansas City.
Robertson, P.K., Huges, J.M. O., Camanella, R.G., Brown, P., & KcKeown, S. (1986). Design of
laterally loaded piles using the pressuremeter. The Pressuremeter and Its Marine Applications:
2nd Int Symp., ASTM STP 950, 443–457.
Rollins, K.M., Peterson, K.T., & Weaver, T.J. (1998). Lateral load behavior of full-scale pile group
in clay. J. Geotech. Geoenvir. Engrg., ASCE, 124(6), 468–478.
Rosenberg, P., & Journeaux, N.L. (1976). Friction and bearing tests on bedrock for high capacity
socket design. Can. Geotech. J., Ottawa, Canada, 13(3), 324–33.
Rowe, R.K., & Armitage, H.H. (1984). The Design of Piles Socketed into Weak Rock. Faculty of
Engineering Science, The University of Western Ontario, London, Ont, Research Report
GEOT11–84.
Rowe, R.K., & Armitage, H.H. (1987a). Theoretical solutions for the axial deformation of drilled
shafts in rock. Can. Geotech. J., Ottawa, Canada, 24, 114–125.
Rowe, R.K., & Armitage, H.H. (1987b). A design method for drilled piers in soft rock. Can.
Geotech. J., Ottawa, Canada, 24,126–142.
Rowe, R.K., & Pells, P.J. N. (1980). A theoretical study of pile-rock socket behavior. Proc. Int
Conf on Structural Found on Rock, Sydney, 1,253–264.
Rutledge, J.C., & Preston, R.L. (1978). Experience with engineering classifications of rock. Proc.
Int. Tunneling Symp., Tokyo, A3.1-A3.7.
Schmertmann, J.H. (1978). Guideline for Cone Penetration Test Performance and Design. Federal
Highway Administration, U.S. Dept. of Transportation, Washington, DC.
Schmertmann, J.H., & Hayes, J.A. (1997). The Osterberg cell and bored pile testing—A symbiosis.
3rd Int. Geotech. Engrg. Conf., Cairo University, Egypt.
Seidel, J.P., & Haberfield, C.M. (1995). The axial capacity of pile sockets in rocks and hard soils.
Ground Engrg., 28(2), 33–38.
Seidel, J.P., & Collingwood, B. (2001). A new socket roughness factor for prediction of rock socket
shaft resistance. Can. Geotech. J., 38, 138–153.
Sen, Z., & Essa, E.A. (1992). Rock quality charts for log-normally distributed block sizes. Int. J.
Rock Mech. Min. Sci. & Geomech. Abstr., 29(1), 1–12.
Sen, Z., & Kazi, A. (1984). Discontinuity spacing and RQD estimates from finite length scanlines.
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 21(4), 203–212.
Serafim, J.L., & Pereira, J.P. (1983). Considerations of the geomechanics classification of
Bieniawski. Proc. Int. Symp. Eng. Geol Underground Constr., Lisbon, 1, II33-II42.
Serrano, A., & Olalla, C. (1996). Allowable bearing capacity of rock foundations using a non-linear
failure criterion. Int J. Rock Mech. Min. Sci. & Geomech. Abstr., 33(4), 327–345.
Shanley, R.J., & Mahtab, M.A. (1976). Delineation and analysis of centers in orientation data.
Mathematical Geology, 8(1), 9–23.
References 420
Sharma, P.V. (1997). Environmental and Engineering Geophysics. Cambridge University Press,
Cambridge, UK.
Sharma, S. (1991). XSTABL—An Integrated Slope Stability Analysis Method for Personal
Computers. Version 4.00. Interactive Software Designs, Inc., Moscow, ID.
Singh, B. (1973). Continuum characterization of jointed rock masses. Part I—The constitutive
equations. Int J. Rock Mech. Min. Sci. & Geomech Abstr., 10, 311–335.
Singh, M., Rao, K.S., & Ramamurthy, T. (2002). Strength and deformational behavior of jointed
rock mass. Rock Mech. and Rock Engrg., 35, 45–64.
Skempton, A.W. (1951). The bearing capacity of clays. Build. Res. Congress, London, Inst. Civ.
Engrg., 1,180.
Souley, M., & Homand, F. (1996). Stability of jointed rock masses evaluated by UDEC with an
extended Saeb-Amadei constitutive law. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 33(3),
233–244.
Sowers, G.F. (1996). Building on Sinkholes—Design and Construction of Foundations in Karst
Terrain. ASCE Press, New York.
Spencer, E.W. (1969). Introduction to the Structure of the Earth. McGraw-Hill, New York.
Steffen, O., et al. (1975). Recent developments in the interpretation of data from joint surveys in
rock masses. 6th Reg. Conf. for Africa on Soil Mech. and Found., II, 17–26.
Stevens, J.B., & Audibert, J.M. E. (1979). Re-examination of p-y curve formulations. Proc.
Offshore Technology Conf., Houston, 1, 379–403.
Stewart, C.L. (1977). Subsurface rock mechanics instrumentation program for demonstration of
shield-type longwall supports at York Canyon, Raton, New Mexico. Proc. 18th U.S. Symp. on
Rock Mech., Keystone, CO, 1C-2, 1–13.
Sun, K. (1994). Laterally loaded piles in elastic media. J. Geotech. Engrg., ASCE, 120(8), 1324–
1344.
Tang, Q. (1995). Load Transfer Mechanisms of Drilled shaft Foundations in Karstic Limestone
Behavior under Working Load. PhD thesis, The University of Tennessee, Knoxville, TN.
Teng, W.C. (1962). Foundation Design. Prentice-Hall Inc., Englewood Cliffs, N.J.
Terzaghi, K. (1946). Rock defects and loads on tunnel supports. Rock Tunneling with Steel
Supports, Youngstown, OH, Eds: R.V. Proctor & T.L. White, 1, 17–99.
Terzaghi, K., & Peck, R.B. (1948). Soil Mechanics in Engineering Practice. Wiley, NY.
Terzaghi, R. (1965). Sources of error in joint surveys. Geotechnique, 5(3), 287–304.
Thone, C.P. (1980). The capacity of piers drilled into rock. Proc. Int. Conf on Structural Found. on
Rock, Sydney, 1,223–233.
To, A. (1997). Lateral Load Capacity of Drilled Shafts in Jointed Rock. MS thesis, Massachusetts
Institute of Technology, Cambridge, MA.
Toh, C.T., Ooi, T. A, Chiu, H. K, Chee, S.K., & Ting, W.N. (1989). Design parameters for bored
piles in a weathered sedimentary fortnation. Proc. 12th Int. Conf, on Soil Mech. and Found.
Engrg., Rio de Janeiro, 2,1073–1078.
Tomlinson, M.K. (1977). Pile Design and Construction Practice. Viewpoint Publications, London.
Tse, R., & Cruden, D.M. (1979). Estimating joint roughness coefficients. Int. J. Rock Mech. Min.
Sci. & Geomech. Abstr., 16(4), 303–307.
Turner, J.P., Sandberg, E., & Chou, N.N. S. (1993). Side resistance of drilled shafts in the Denver
and Pierre Formations. Design and Performance of Deep Foundations: Piles and Piers in Soil
and Soft Rock, Geotechnical Special Publication No. 38, 245–259.
Vallabhan, C.V. G., & Alikhanlou, F. (1982). Short rigid piers in clays. J. Geotech. Engrg., ASCE,
108(10), 1255–1272.
Verruijt, A., & Kooijman, A.P. (1989). Laterally loaded piles in a layered elastic medium.
Vijayvergiya, V.N. (1977). Load-movement characteristics of piles. Proc. 4th Symp. of Waterways,
Port, Coastal and Ocean Division, ASCE, Long Beach, CA, 269–284.
References 421
Villaescusa, E. (1993). Statistical modeling of rock jointing. Proc. Conf. on Probabilistic Methods
in Geotech. Engrg., Lanberra, Australia, 221–231.
Villaescusa, E., & Brown, E.T. (1992). Maximum likelihood estimation of joint size from trace
length measurements. Rock Mech. and Rock Engrg., 25, 67–87.
Voight, B. (1968). On the functional classification of rocks for engineering purposes. Int. Symp. on
Rock Mech, Madrid, 131–135.
Voss, R. (1988). Fractals in nature. The Science of Fractal Images, Eds: H. Peitgen & D. Saupe,
Springer, New York, 21–69.
Wakai, A., Gose, S., & Ugai, K. (1999). 3-D Elasto-plastic finite element analysis of pile
foundations subjectedto lateral loading. Soils and Found., 39(1), 97–111.
Wallis, P.F., & King, M.S. (1980). Discontinuity spacing in a crystalline rock. Int. J. Rock Mech.
Min. Sci. & Geomech. Abstr., 17, 63–66.
Wang, S. (1992). Fundamental Studies of the Deformability and Strength of Jointed Rock Masses
at Three Dimensional Level. PhD Dissertation, University of Arizona, Tucson.
Wang, S., & Kulatilake, P.H. S.W. (1993). Linking between joint geometry models and a distinct
element method in three dimensions to perform stress analyses in rock masses containing finite
size joints. Soils and Found., 33(4), 88–98.
Warburton, P.M. (1980a). A stereological interpretation of joint trace data. Int. J. Rock Mech. Min.
Sci. & Geomech. Abstr., 17, 181–190.
Warburton, P.M. (1980b). Stereological interpretation of joint trace data: Influence of joint shape
and implications for geological surveys. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 17,
305–316.
Wathugala, D.N. (1991). Stochastic Three Dimensional Joint Geometry Modeling and Verification.
Ph.D. Dissertation, University of Arizona, Tucson.
Wathugala, D.N., Kulatilake, P.H. S.W., Wathugala, G.W., & Stephansson, O. (1990). A general
procedure to correct sampling bias on joint orientation using a vector approach. Computers and
Geomech., 10, 1–31.
Webb, D.L. (1976). The behavior of bored piles in weathered diabase. Geotechnique, 26(1), 63–72.
Webber, I.P., & Gowans, D. (1996). A new core orientation device. Proc. Int. Conf. on Advances in
Site Investigation Practice, London, 306–311.
Whitaker, T. (1975). The Design of Pile Foundations. 2nd edition, Pergamon, Oxford.
Wickham, G.E., Tiedemann, H.R., & Skinner, E.H. (1972). Support determination based on
geologic predictions. Proc. North American Rapid Excav. Tunneling Conf., Chicago, Eds: K. S.
Lane & L.A. Garfield, 43–46.
Williams, A.F. (1980). The Design and Performance of Piles Socketed into Weak Rock. Ph.D.
dissertation, Monash University, Clayton, Victoria, Australia.
Williams, A.F., Johnston, I.W., & Donald, I.B. (1980). The design of socketed piles in weak rock.
Proc. Int Conf. on Structural Found. on Rock, Sydney, 1,327–347.
Williams, A.F., & Pells, P.J. N. (1981). Side resistance of rock sockets in sandstone, mudstone, and
shale. Can. Geotech J., Ottawa, Canada, 18, 502–513.
Wilson, E.L. (1977). Finite Elements for Foundations, Joints and Fluids. Finite Elements in
Geomech., Ed. G. Gudehus, John Wiley & Sons, NY, 319–329.
Wilson, L.C. (1976). Tests of bored and driven piles in cretaceous mudstone at port Elizabeth,
South Africa. Geotechnique, 26(1), 5–12.
Wilson, W.L., & Beck, B.F. (1988). Evaluating sinkhole hazards in mantled karst terrane.
Geotechnical Aspects of Karst Terrains: Exploration, Foundation Design and Performance, and
Remedial Measures, GSP No. 14, Ed.: N. Sitar, ASCE, New York, NY, 1–24.
Winterkorn, H.F., & Fang, H.-F.(1975). Foundation Engineering Handbook. Van Nostrand
Reinhols, New York, 601–615.
Wright, S.G. (1991). UTEXAS3—A Computer Program for Slope Stability Calculations. Software
Manual. Shinoak Software. Texas.
Wyllie, D.C. (1999). Foundations on Rock (2nd ed.). E & FN SPON.
References 422
Yoshinaka, R., & Yambe, T (1986). Joint stiffness and the deformation behavior of discontinuous
rock. Int. J. Rock Mech. Min. Sci. & Geomech Abstr., 23(1), 19–28.
Yu, X., & Vayssade, B. (1990). Joint profiles and their roughness parameters. Proc. Int. Symp. on
Rock Joints, Loen, Norway, Eds: N. Barton & O. Stephanson, Balkema, Rotterdam, 781–785.
Yudhbir, Lemanza, W., & Prinzl, F. (1983). An empirical failure criterion for rock masses. Proc.
5th Int Cong. on Rock Mech., Melbourne, 1, B1-B8.
Zanbak, C. (1977). Statistical interpretation of discontinuity contour diagrams. Int. J. Rock Mech.
Min. Sci. & Geomech. Abstr., 14, 111–120.
Zhang, L. (1999). Analysis and Design of Drilled Shafts in Rock. PhD thesis, Massachusetts
Institute of Technology, Cambridge, MA.
Zhang, L., & Einstein, H.H. (1998a). End bearing capacity of drilled shafts in rock. J. Geotech
Geoenvir. Engrg., ASCE, 124(7), 574–584.
Zhang, L., & Einstein, H.H. (1998b). Estimating the mean trace length of rock discontinuities. Rock
Mech. and Rock Engrg., 31(4), 217–235.
Zhang, L., & Einstein, H.H. (1998c). Prediction Results of Fitchburg Drilled Shafts for MHD.
Department of Civil and Environmental Engineering, MIT.
Zhang, L., & Einstein, H.H. (2000a). Estimating the intensity of rock discontinuities. Int. J. Rock
Mech. Min. Sci., 37(5), 819–837.
Zhang, L., & Einstein, H.H. (2000b). Estimating the deformation modulus of rock masses. Paciflc
Rocks 2000, Proc. 4th North American Rock Mech. Symp., Seattle, WA, Eds: J. Girard, et al.,
703–708.
Zhang, L., Einstein, H.H., & Dershowitz, W.S. (2002). Stereological relationship between trace
length distribution and size distribution of elliptical discontinuities. Geotechnique, 52(6), 419–
433.
Zhang, L., Ernst, H., & Einstein, H.H. (2000). Nonlinear analysis of laterally loaded rock-socketed
shafts. J. Geotech Geoenvir. Engrg., ASCE, 126(11), 955–968.
Zhang, L., Silva, F.-T., & Grismala, R.F. (2002). Ultimate resistance of laterally loaded piles in
cohesionless soils. Deep Foundations 2002, Proc. Int. Deep Found. Cong. 2002, GSP No. 116,
ASCE, Eds: M.W. O’Neill & F.C. Townsend, 2,1364–1375.
Zhu, W., & Wang, P. (1993). Finite element analysis of jointed rock masses and engineering
application. Int. J. Rock Mech Min. Sci. & Geomech. Abstr., 30(5), 537–544.
Zienkiewicz, O.C., et al. (1970). Analysis of nonlinear problems with particular reference to jointed
rock systems. Proc. 2nd Int. Cong. on Rock Mech., Belgrade, 3, 501–509.
Zongqi, S., & Xu, F. (1990). Study of rock joint surface feature and its classification. Proc. Int.
Symp. on Rock Joints, Loen, Norway, Eds: N.Barton & O.Stephanson, Balkema, Rotterdam,
101–107.
Index
acoustic televiewer 162

acoustic wave 165
adhesion factor 192, 196, 202
agglomerate 124
air 165, 166
allowable
bearing pressure 204, 206, 208, 210
concrete stress 189, 190
design load 6
design stress 190
deformation 8
differential deformation 8
end bearing resistance 206
stress design 6
working load 6
alpite 143
amphibolite 13, 16, 17, 124, 130
andesite 11, 17, 124, 130, 131
angle
basic friction 77, 78, 82
dilation 142, 143, 245
dip 36
internal friction 77, 132
plunge 54–56
residual friction 77
roughness 77
trend 54–56
anhydrite 124, 131, 315
anisotropy 71, 93, 132, 150, 298
anorthosite 16
aperture of discontinuity 46
aplite 84
apparent cohesion 77
area of discontinuity 38
argillite 17, 199
asperity 82, 84, 195, 196
augers 1
augite 10, 11
axial deformation influence factor 235, 238
axial displacement 227–250
axial load capacity 189–225
single shafts 189–221
shaft groups 221–225
Index 424
axial loading test

compressive 328–344
uplift 344, 345
Barton model 78, 142

basalt 11, 13, 15, 17, 18, 77, 124, 131, 144, 166
basic friction angle 77, 78, 82
beam 264, 287
bearing capacity 204–220
bedding plane 34
bending moment 251, 264, 275, 282, 284, 347–352
bias in sampling
frequency 154
orientation 54, 156
spacing 154
trace length 58–63, 155
bilinear shear strength model 77
biotite 10, 11
body wave 165
bored piles 1
borehole camera 162
borehole core
logging 158
sampling 157
borehole dilatometer test 180–183
borehole jack test 182, 183
borehole periscope 162
boring 157
boundary condition 261, 280, 284, 305
boundary element method 270
breccia 12, 124, 218, 233
bridge 3, 307
bridge foundation 3, 307
buckling 187, 329
building codes 191, 210
caissons 1
calcite 10, 12
capacity
axial 189–225
lateral 251–262
carbonatite 17
casing 321, 323
cast-in-place pile 1
cave 321, 323
cavity 6, 7, 315, 316, 318, 324
chalk 13, 16, 124
charnockite 131
chert 13, 128
Christensen-Huegel method 159
circular disks 42
Index 425
circular failure 307, 308, 314

classification
engineering 14, 15
geological 10–13
intact rock 9–14
rock mass 14–31
weathering 13
clay 131, 166, 175, 248
clay core barrel method 159
clay seam 5
clayshale 217, 218
claystone 12, 13, 17, 124, 131, 146
cleavage 35
closed circuit television (CCTV) 162
coal 12, 124, 146
cohesion 77, 127, 132, 141, 179, 245
compressive wave 18, 165
computer programs
COMP624P 264
LPILE 264
UTESAS3 314
XSTABL 314
concrete strength 190, 251–254
conglomerate 12, 17, 124
constant rate of penetration 340, 345
constant rate of uplift 345
continuum approach 234, 269
Iinear 234, 270
nonlinear 245, 278
core
boring 157
logging 158
orientation 159
Coulomb model 77, 141
Craelius method 159
cross hole method 171
cyclic loading 340
dacite 11, 124

defects 318, 352
deflection 263, 270
deflection influence factor 270, 271
deflection ratio 300, 302
deformability of discontinuities
normal stiffness 73–76
shear stiffness 73–76
deformability of rock mass
empirical methods 87–93
equivalent continuum approach 93–116
scale effect 149
deformation
Index 426
allowable 8
allowable differential 8
deformation modulus 86–116, 149
density 14, 16, 165
design load 6
design stress 190
diabase 13, 15–18, 130, 217, 233
diamond drilling 157
dielectric property 175
dilatancy 245, 258
dilation 196, 204, 245
dilation law 245
dioritell, 16, 17, 124, 131
dip direction 36
direct shear test 176, 178
direction cosines 54
discontinuity
aperture 46, 47
apparent cohesion 77
basic friction angle 77
circular disks 42
cohesion 77
deformability 73–76
ellipse 43
filling 47
frequency 33, 36, 57
internal friction angle 77
orientation 20, 33, 36, 48–56, 60, 64, 67, 68, 159–161
persistence 37–42
persistence ratio 38–42
roughness 44–46
roughness coefficient 78–85
roughness profile 79–82
sampling 153–164
set 54
shape 42–44
size 37, 63–68
spacing 20, 36, 57
stiffness 73–75
strength 73, 76–85
trace length 58–63
wall compressive strength 78, 82–84
discontinuum method 261, 262
discrete element method (DEM) 262, 298
displacement
axial 227–250
lateral 263–305
normal 73–76, 142, 245
shear 245
dolerite 11, 124, 128, 130
dolomite 10, 12, 15, 16, 130, 131, 315
dolostone 17, 166
Index 427
drilled caissons 1
drilled piers 1
drilled shafts 1, 189, 227, 251, 263, 307, 315, 327
drilling
clay core barrel 159
diamond 157
directional 157
integral sampling 159
large diameter 157
dunite 11, 16
durability 6, 8
eclogite 16
effective roughness angle 77
elastic continuum 234, 269
elastic modulus 14, 17
electrical resistivity 174, 175
ellipse 43
empirical relations
deformation modulus 87–93
end bearing capacity 209–220
side shear resistance 192–198
empirical rock mass strength criterion 123–131
end bearing resistance 191, 204–220
engineering geology 34
equal-angle projection 50
equal-area projection 50
equivalent continuum approach
deformability 93–116
strength 132–139
equivalent pier 248–250
exploratory tunnel 164
exponential distribution 57, 59, 64, 65, 67, 68
factor of safety 6, 307, 309, 313

failure criterion 139
failure type
circular 307, 308, 314
planar 307–312
toppling 307, 308, 314
wedge 307, 308, 314
fault 34
filling 47
finite difference method 282
finite element method (FEM) 272, 298, 305
flat jack test 187
foliation 35
fracture tensor 68, 69
frequency of discontinuities 33, 36, 57
friction angle 77, 78, 82, 127, 134, 179
Index 428
gabbro 11, 13, 16–18, 124, 130, 144

Gamma distribution 59, 64, 65, 67, 68
gauges 331, 336–339, 347, 348
geologic structure 157
geological strength index (GSI) 29–31, 125, 126
geology 151, 152, 164
geomechanics classification 20
geophysical exploration 164–174
gneiss 12, 13, 15–18, 77, 84, 87, 124, 130, 143
Goodman jack 182
granite 11, 13, 15–18, 77, 84, 124, 128, 130, 131, 143, 144
granodiorite 11, 13, 16, 124
granulite 16
great circle 48–50
greywacke 124
gritstone 12
ground penetration radar 175, 318
groundwater 5, 6
group
block 222, 223, 249, 261
capacity 221–225, 261
displacement 248–250, 300–305
efficiency 221–224
interaction 221, 249, 303
settlement 248
grout 171
gypstone 124
gypsum 217, 233, 315
halite 315
hardness 10
hardpan 217
hematite 10
hemispherical projection 48
equal-angle 50
equal-area 50
great circle 48, 49
histogram 57
Hoek-Brown strength criterion 123, 211, 258
Hong Kong 206, 209
hornblende 10–12
hornfels 12, 84, 124, 143
hydraulic jack 329, 337, 346
in-plane failure 40–42, 310, 311

in situ shear tset 178, 179
in situ test 177–188
infinite element 247, 248
influence factor
axial deformation 235, 238
deflection 270, 271
Index 429
rotation 270–272
intact rock 9
integral sampling method 195
integrity test
cross-hole sonic logging test 352, 353
low strain integrity test using PIT 352, 353
interface cohesion 245
interface friction 245
internal friction angle 77, 132
ironstone 12
joint 35
joint element
nodal displacement 117
relative nodal displacement 120
joint roughness coefficient (JRC) 78–85, 143
kaolinite 10, 12, 26

karst 317
karst area 317
karst terrain 315
karstic formations 6, 7, 315, 316, 318, 324
kinematic analysis 262
kinetic analysis 262
laboratory testing 175–177

large-diameter boring 162
lateral loading test 345–352
lateral resistance 254–262
limestone 12, 13, 15–17, 77, 87, 124, 128, 130, 131, 144, 166, 175, 194, 218, 222, 233, 315
limit equilibrium approach 262
liparite 131
load distribution 343, 344
load-transfer method
linear 228
nonlinear 231
loading test
compression 328–344
constant rate of penetration 340, 345
constant rate of uplift 345
lateral 345–352
maintained load 339
uplift 123, 124
loess 131
logging
borehole core 158
scanline 153
lognormal distribution 59, 64, 65, 67, 68
low-angle-transition 41, 42
lower hemisphere projection 49, 50
LVDT 339
Index 430
maintained load test 339

marble 12, 13, 15–18, 124, 130, 131, 144
marl 13, 194, 218, 233
marlstone 17
mean
frequency 57
orientation 55, 56
spacing 57
trace length 59–63
migmatite 124
modulus
deformation 86
elastic 86, 165
initial tangent 85, 86
recovery 86
shear 165
Young’s 165
monozonite 17
mudstone 12, 128, 130, 131, 190, 194, 198, 217
muscovite 10, 11
Mustran cells 338
mylonites 124
norite 16, 124, 128, 130, 131, 144

normal
displacement 73–76, 142, 245
stiffness 73–76
stress 74–79, 141–143
numerical methods
discrete element method (DEM) 262, 298
finite difference method 282
finite element method (FEM) 272, 298, 305
obsidian 124
olivine 10, 11
orientation of discontinuities
dip 36
distribution 56
mean 55, 56
plunge 54–56
sampling bias 55
strike 36
trend 54–56
polar stereonet 51, 52
orthoclase 11
orthoclase feldspar 10
orthoclase porphyries 11
Osterberg cell 330–335, 346, 347
Index 431
P-wave 165
P-wave velocity 29, 165, 166
p-y curve 263, 264, 266–269, 303, 350–352
pendulum orientation method 159
peridotite 11, 13, 16,
persistence of discontinuities 37–42
persistence ratio 38–42
photographic mapping 156
phyllite 12, 13, 17, 124, 294
picrite 11
piers 1
piles 1
Pile Integrity Tester (PIT) 352, 353
plagioclase 11
plagioclase feldspar 10
plagioclase porphyries 11
planar sliding failure 307–312
plane strain 245
plate bearing test 184–186
plunge 54–56
point load index 13, 209
point load test 13, 176
Poisson’s ratio 14, 18, 165
polar stereonet 51, 52
porosity 176
potash 13
pressure cells 339
pressure wave 165
pressuremeter 178, 206
primary wave 165
probability distribution
discontinuity orientation 56
discontinuity spacing 57
discontinuity trace length 59
PVC casing 171
pyroxenite 16
Q-system 20, 23–29

quartz 10–13, 16, 26
quartz diorite 17, 128, 130, 144
quartz monozonite 17
quartzite 12, 13, 15, 17, 18, 124, 128, 130, 131, 294
radial
displacement 245
radial jacking 188
stress 245
Rayleigh wave 165, 171
reconnaissance 151, 152, 174
reinforcement 189, 190, 253, 254
residual friction angle 77
Index 432
resistance
end bearing 191, 204–220
lateral 254–262
side shear 191–204
resisting force 308
resistivity survey 174
rhyolite 11, 13, 124, 130
RMR-Q relation 29
rock
bridge 40–42, 310, 311
core boring 157
igneous 10, 11, 166, 175
metamorphic 10, 11, 166, 175
sedimentary 10, 11, 166
weathered 166
weathering 11
rock mass 14
rock mass rating (RMR) 20–23, 29, 31, 90–92, 125
rock quality designation (RQD) 14, 16, 18–21, 87–91
rotation 270
rotation influence factor 270–272
roughness
angle 77
coefficient 78–85, 143
discontinuity 44–46
factor 195, 196
height 195, 198, 200, 201
interface 196
number 23
profile 79–82, 195, 199
shaft 195–200
S-wave 165
salt 13, 131, 315
sampling
bias 54, 58, 59, 154–156
photographic mapping 156
scanline 153–155
window 155, 156
sand 16, 166, 175, 248
sandstone 5, 12, 13, 15–18, 43, 77, 87, 124, 128, 130, 131, 166, 175, 177, 192, 194, 199, 203, 210,
217, 218, 291–294
scale effect 82–84, 144–149
scanline
circular 155
straight 153–155
scanline sampling 153–155
schist 12, 13, 15–18, 124, 146
scour 189, 295
seam 319–322
seismic survey 165
Index 433
seismic wave 165

serpentinite 11
settlement ratio 248
shaft group 221, 248, 261, 300
shaft resistance coefficient (SRC) 198–200
shale 5, 12, 13, 15–18, 84, 128, 130, 131, 143, 146, 166, 175, 192, 194, 199, 217, 218, 291–294
shape of discontinuities 42–44
shear
displacement 73–76, 142, 245
modulus 165
resistance 191–204
stiffness 73–76
wave 165
wave velocity 165
side shear resistance 191–204
siltstone 3, 4, 13, 17, 18, 77, 124, 128, 130, 131, 192, 210, 218, 220, 277
sink hole 315, 316, 324
sister bars 338
size of discontinuities 37, 63–68
slate 12, 16, 17, 77, 84, 124, 130, 131, 143
sliding force 307
slope
sliding 307–314
stability 307–314
soapstone 84
socketed shaft 1
socket 1
soluable rocks 315
sonic logging 352, 353
sound velocity 352
spacing of discontinuities 20, 36, 57
spacing of drilled shafts 248, 303, 329, 330, 346
spacing ratio 148
standard penetration test (SPT) 191, 192
Statnamic loading test 335, 346
steel 166
stereographic projection 48–51
stiffness
normal 73–76
shear 73–76
strain gauges 332, 347, 348
strength of discontinuities
Barton model 78, 142
bilinear shear strength model 77
Coulomb model 77, 141
strength of rock mass
Bieniawski-Yudhbir criterion 127
equivalent continuum approach 132
Hoek-Brown criterion 123
Johnston criterion 128
Ramamurthy criterion 130
scale effect 144
Index 434
strike 36
structural geology 34, 151
subgrade reaction approach
linear 265
nonlinear 266
syenite 16, 17
t-z curve 228–234, 249, 343, 344

tangent modulus 85, 86
telltales 338
tensile strength 40, 311
tensor 68
test
axial 328–344
borehole dilatometer 180–181
borehole jack 182, 183
flat jack 187
in situ 177–188
in situ shear 178, 179
integrity 352, 353
laboratory 175–177
lateral 345–352
pits 163
plate bearing 184–186
radial jacking 188
tilt 81
uplift 344, 345
thin-layer element 121
tiff 13
till 217
tilt angle 79, 81
tilt test 81
toppling failure 307, 308, 314
trace length
mean 59–63
probability distribution 59, 62–68
sampling bias 58, 59
trenches 163
trend 54–56
truncation 69, 60
tuff 124, 128, 131
unconfined compressive strength 14, 17, 20, 91–93, 123, 127–131, 144, 176–178, 192–203, 209–
211, 215–220
uniaxial compressive strength, see unconfined compressive strength
universal distinct element code (UDEC) 298
uplift
load 222–225
test 344, 345
vector 54, 68
Index 435
volumetric joint count 19
water 166
weathering 11, 315
wedge sliding failure 307, 308, 314
window sampling 155, 156,
yield
deflection factor 278
rotation factor 278
strength 190, 251–253
yielding 264, 278
Young’s modulus 165

Pile PDF

Uploaded by

Copyright:

Available Formats

Pile PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pile PDF

Uploaded by

Copyright:

Available Formats

What are drilled shafts and how are they constructed?

What are drilled shafts and how are they constructed?

How did drilled shaft construction develop historically?

How did drilled shaft construction develop historically?

Contents

1.1 DEFINITION OF DRILLED SHAFTS

A drilled shaft is a deep foundation that is constructed by placing concrete in an

1.2 HISTORICAL DEVELOPMENT OF DRILLED SHAFTS

Fig. 1.1 A drilled shaft socketed into

1.3 USE OF DRILLED SHAFTS

An application example of drilled shafts in rock

1.4 CHARACTERISTICS OF DRILLED SHAFTS IN ROCK

1.4.1 Effect of discontinuities

Fig. 1.2 Effect of discontinuities on the

1.4.2 Effect of groundwater

1.4.3 Effect of karstic formations

1.5 CONSIDERATIONS IN THE DESIGN OF DRILLED SHAFTS

Fig. 1.3 Failure modes of drilled shafts

2.2 INTACT ROCK

2.2.1 Geological classification

(a) Rock-forming minerals

(b) Elementary rock classification

(c) Weathering of rock

Major mineral Quartz, orthoclase, Quartz, orthoclase, Calcium- Calcium-

Table 2.3 Classification of sedimentary rocks.

Table 2.4 Classification of metamorphic rocks.

Quartzite Fined grained Quartz

Table 2.5 Degree of weathering of intact rocks.

Table 2.6 Engineering classification of rock by

2.2.2 Engineering classification

2.2.3 Typical values of intact rock properties

2.3 ROCK MASS

2.3.1 Rock quality designation (RQD)

(a) Direct method for determining RQD

Fig. 2.1 Engineering classification of

Table 2.7 Typical values of density of intact rocks

Table 2.8 Typical range of unconfined compressive

Table 2.9 Typical values of elastic modulus of

Diabase 7 7 104 69.0 88.3 12.3

Table 2.10 Typical values of Poisson’s ratio of

Table 2.11 Correlation between RQD and rock

(b) Indirect methods for determining RQD

Fig. 2.2 Procedure for measurement

2.3.2 Rock mass rating (RMR)

2.3.3 Rock mass quality (Q)

Table 2.12 Geomechanics classification of jointed

B. Rating adjustment for joint orientations

Table 2.13 Guidelines for classifying discontinuity

Description and None Hard Hard Soft Soft filling

Table 2.14 Ratings for discontinuity orientations for

• Rock Quality Designation (RQD)

Table 2.15 The Q-system and associated parameters

(ii) Take RQD to be nearest 5%

impermeable filling, i.e. quartz or epidote

rock and clay (see G, H. J for

surrounding rock (any

Table 2.16 Classification of rock mass based on Q-

RMR=A loge Q+B

where A is typically in the range 5–15, and B in the range 35–60.